diff --git a/.capabilities/data/quality/analyze-sparse-observations.capability.yaml b/.capabilities/data/quality/analyze-sparse-observations.capability.yaml
index e88c986..651f8f2 100644
--- a/.capabilities/data/quality/analyze-sparse-observations.capability.yaml
+++ b/.capabilities/data/quality/analyze-sparse-observations.capability.yaml
@@ -21,6 +21,8 @@ guidance:
   - Default missingness is unknown. A modeled expectation is never a measurement.
   - Caller business severity, incident state and notifications remain outside
     the engine.
+  - Optional event-identity-v1 IDs retain simultaneous distinct calls; only a
+    separately asserted trusted sequence permits ordering their measurements.
 planning:
   story_map:
     backbone: Monitor while periods are still open
diff --git a/.capabilities/data/quality/identify-distinct-events.capability.yaml b/.capabilities/data/quality/identify-distinct-events.capability.yaml
new file mode 100644
index 0000000..fe815b7
--- /dev/null
+++ b/.capabilities/data/quality/identify-distinct-events.capability.yaml
@@ -0,0 +1,363 @@
+title: Inspect distinct events sharing an original timestamp
+status: implemented
+summary: Callers supply stable logical event IDs so distinct comparable calls at
+  one timestamp retain separate values and evidence without inventing
+  chronology.
+intent: Make real per-call evidence consumable while protecting references,
+  replay and pattern interpretation from retry duplication and arbitrary tie
+  order.
+acceptance:
+  - Identified event-index sources retain every distinct event and its original
+    time, reject repeated logical IDs and preserve calendar and legacy
+    semantics.
+  - Without a trusted sequence, whole timestamp cohorts share prior-time point
+    references and ties cannot establish change, Nelson or CUSUM continuity.
+  - Explicit trusted sequence metadata establishes a validated event order, not
+    independence, elapsed cadence or operational authority.
+  - Event-specific evidence and case identities survive deterministic replay;
+    availability and identified corrections retain as-of provenance and bounded
+    results.
+guidance:
+  - Callers own identity authenticity, retry deduplication, compatible
+    populations and any assertion of trustworthy sequence. Sorting an ID
+    establishes no cause.
+  - Cases and overlapping findings are inspectable numerical evidence;
+    applications own incident deduplication, business severity and
+    notifications.
+#
+#
+agent:
+  # -----------------------------
+  # machine managed agent metadata
+  # review all capabilities and save evidence: capabilitykit review
+  # review this capability and save evidence: capabilitykit review data/quality/identify-distinct-events
+  # run deterministic review only: capabilitykit review data/quality/identify-distinct-events --deterministic-only
+  # ask an agent and save review evidence: capabilitykit review data/quality/identify-distinct-events --agent codex --arg exec --handoff stdin
+  # validate saved agent output without writing: capabilitykit review-result data/quality/identify-distinct-events --input review.json
+  # -----------------------------
+  depends_on:
+    - data/inputs/accept-analysis-requests
+    - data/quality/analyze-sparse-observations
+    - analysis/detection/identify-residual-patterns
+  implementation:
+    references:
+      - src/anomalyzer/contracts.py
+      - src/anomalyzer/events.py
+      - src/anomalyzer/sparse.py
+      - src/anomalyzer/orchestrator.py
+      - src/anomalyzer/alignment.py
+      - src/anomalyzer/relationships.py
+      - src/anomalyzer/cases.py
+      - src/anomalyzer/mcp_views.py
+      - tests/test_events.py
+      - docs/sparse-data.md
+      - examples/sparse_examples.py
+      - examples/render_readme_progression.py
+  verification:
+    automated:
+      - id: identity-and-retries
+        description: Retain distinct tied calls and reject duplicate logical identities.
+        command: python -m pytest -q tests/test_events.py -k "identity or repeat or
+          distinct or reordering"
+      - id: honest-order
+        description: Keep point/range evidence and require actual declared order for
+          tied sequences.
+        command: python -m pytest -q tests/test_events.py -k "cohort or ties or trusted
+          or reset or lag"
+      - id: reproducible-replay
+        description: Preserve prefix, as-of, addressed corrections and case revisions.
+        command: python -m pytest -q tests/test_events.py -k "prefix or asof or
+          correction or revision or reference"
+      - id: public-consumers
+        description: Validate schemas, Python, CLI, MCP retrieval and bounded response
+          growth.
+        command: python -m pytest -q tests/test_events.py -k "python_schema or bounded"
+    manual:
+      - Inspect progression figures 23–24 and the API guide for exact original
+        times, layout versus trusted order, retry responsibility and overlapping
+        evidence limits.
+  review:
+    depth: partial
+    source: deterministic-assessment
+    intent_summary: Implementation evidence synchronized from deterministic
+      assessment for data/quality/identify-distinct-events.
+    done: false
+    criteria:
+      - criterion: Identified event-index sources retain every distinct event and its
+          original time, reject repeated logical IDs and preserve calendar and
+          legacy semantics.
+        status: partial
+        evidence:
+          - src/anomalyzer/contracts.py:81
+          - src/anomalyzer/contracts.py:118
+          - src/anomalyzer/contracts.py:192
+          - src/anomalyzer/contracts.py:193
+          - src/anomalyzer/contracts.py:241
+          - src/anomalyzer/events.py:1
+          - src/anomalyzer/events.py:14
+          - src/anomalyzer/events.py:22
+          - src/anomalyzer/events.py:23
+          - src/anomalyzer/events.py:26
+          - src/anomalyzer/sparse.py:4
+          - src/anomalyzer/sparse.py:23
+          - src/anomalyzer/sparse.py:46
+          - src/anomalyzer/sparse.py:60
+          - src/anomalyzer/sparse.py:62
+          - src/anomalyzer/orchestrator.py:21
+          - src/anomalyzer/orchestrator.py:84
+          - src/anomalyzer/orchestrator.py:90
+          - src/anomalyzer/orchestrator.py:94
+          - src/anomalyzer/orchestrator.py:100
+          - src/anomalyzer/alignment.py:14
+          - src/anomalyzer/alignment.py:33
+          - src/anomalyzer/alignment.py:46
+          - src/anomalyzer/alignment.py:47
+          - src/anomalyzer/alignment.py:48
+          - src/anomalyzer/relationships.py:7
+          - src/anomalyzer/relationships.py:19
+          - src/anomalyzer/relationships.py:33
+          - src/anomalyzer/relationships.py:34
+          - src/anomalyzer/relationships.py:35
+          - src/anomalyzer/cases.py:16
+          - src/anomalyzer/cases.py:18
+          - src/anomalyzer/cases.py:20
+          - src/anomalyzer/cases.py:22
+          - src/anomalyzer/cases.py:24
+          - src/anomalyzer/mcp_views.py:68
+          - src/anomalyzer/mcp_views.py:71
+          - src/anomalyzer/mcp_views.py:76
+          - src/anomalyzer/mcp_views.py:77
+          - src/anomalyzer/mcp_views.py:89
+          - tests/test_events.py:1
+          - tests/test_events.py:22
+          - tests/test_events.py:24
+          - tests/test_events.py:25
+          - tests/test_events.py:41
+          - docs/sparse-data.md:50
+          - docs/sparse-data.md:73
+          - docs/sparse-data.md:74
+          - docs/sparse-data.md:77
+          - docs/sparse-data.md:83
+          - examples/sparse_examples.py:36
+          - examples/sparse_examples.py:57
+          - examples/sparse_examples.py:65
+          - examples/sparse_examples.py:66
+          - examples/sparse_examples.py:69
+          - examples/render_readme_progression.py:141
+          - examples/render_readme_progression.py:301
+          - examples/render_readme_progression.py:303
+          - examples/render_readme_progression.py:373
+          - examples/render_readme_progression.py:413
+        notes: Related implementation text was found, but deterministic matching cannot
+          prove the behavior. Review the cited evidence. If it is correct, save
+          agent.review evidence or make the criterion/reference wording more
+          explicit.
+      - criterion: Without a trusted sequence, whole timestamp cohorts share prior-time
+          point references and ties cannot establish change, Nelson or CUSUM
+          continuity.
+        status: partial
+        evidence:
+          - src/anomalyzer/contracts.py:54
+          - src/anomalyzer/contracts.py:145
+          - src/anomalyzer/contracts.py:246
+          - src/anomalyzer/contracts.py:255
+          - src/anomalyzer/contracts.py:256
+          - src/anomalyzer/events.py:26
+          - src/anomalyzer/events.py:27
+          - src/anomalyzer/events.py:28
+          - src/anomalyzer/events.py:37
+          - src/anomalyzer/events.py:46
+          - src/anomalyzer/sparse.py:1
+          - src/anomalyzer/sparse.py:23
+          - src/anomalyzer/sparse.py:49
+          - src/anomalyzer/sparse.py:54
+          - src/anomalyzer/sparse.py:60
+          - src/anomalyzer/orchestrator.py:11
+          - src/anomalyzer/orchestrator.py:39
+          - src/anomalyzer/orchestrator.py:130
+          - src/anomalyzer/orchestrator.py:144
+          - src/anomalyzer/orchestrator.py:170
+          - src/anomalyzer/alignment.py:62
+          - src/anomalyzer/relationships.py:38
+          - src/anomalyzer/relationships.py:44
+          - src/anomalyzer/relationships.py:164
+          - src/anomalyzer/relationships.py:166
+          - src/anomalyzer/relationships.py:168
+          - src/anomalyzer/cases.py:11
+          - src/anomalyzer/cases.py:25
+          - src/anomalyzer/cases.py:31
+          - src/anomalyzer/cases.py:44
+          - src/anomalyzer/mcp_views.py:76
+          - src/anomalyzer/mcp_views.py:188
+          - src/anomalyzer/mcp_views.py:304
+          - src/anomalyzer/mcp_views.py:322
+          - src/anomalyzer/mcp_views.py:326
+          - tests/test_events.py:18
+          - tests/test_events.py:23
+          - tests/test_events.py:30
+          - tests/test_events.py:47
+          - tests/test_events.py:56
+          - docs/sparse-data.md:31
+          - docs/sparse-data.md:88
+          - docs/sparse-data.md:93
+          - docs/sparse-data.md:106
+          - docs/sparse-data.md:111
+          - examples/sparse_examples.py:65
+          - examples/sparse_examples.py:67
+          - examples/sparse_examples.py:74
+          - examples/sparse_examples.py:81
+          - examples/sparse_examples.py:83
+          - examples/render_readme_progression.py:346
+          - examples/render_readme_progression.py:349
+          - examples/render_readme_progression.py:352
+          - examples/render_readme_progression.py:477
+          - examples/render_readme_progression.py:542
+        notes: Related implementation text was found, but deterministic matching cannot
+          prove the behavior. Review the cited evidence. If it is correct, save
+          agent.review evidence or make the criterion/reference wording more
+          explicit.
+      - criterion: Explicit trusted sequence metadata establishes a validated event
+          order, not independence, elapsed cadence or operational authority.
+        status: partial
+        evidence:
+          - src/anomalyzer/contracts.py:246
+          - src/anomalyzer/contracts.py:255
+          - src/anomalyzer/contracts.py:256
+          - src/anomalyzer/contracts.py:268
+          - src/anomalyzer/contracts.py:296
+          - src/anomalyzer/events.py:1
+          - src/anomalyzer/events.py:50
+          - src/anomalyzer/events.py:71
+          - src/anomalyzer/events.py:80
+          - src/anomalyzer/events.py:84
+          - src/anomalyzer/sparse.py:46
+          - src/anomalyzer/sparse.py:62
+          - src/anomalyzer/sparse.py:93
+          - src/anomalyzer/sparse.py:170
+          - src/anomalyzer/sparse.py:182
+          - src/anomalyzer/orchestrator.py:24
+          - src/anomalyzer/orchestrator.py:142
+          - src/anomalyzer/orchestrator.py:193
+          - src/anomalyzer/orchestrator.py:209
+          - src/anomalyzer/orchestrator.py:213
+          - src/anomalyzer/alignment.py:64
+          - src/anomalyzer/alignment.py:66
+          - src/anomalyzer/alignment.py:68
+          - src/anomalyzer/relationships.py:38
+          - src/anomalyzer/relationships.py:198
+          - src/anomalyzer/relationships.py:205
+          - src/anomalyzer/cases.py:59
+          - src/anomalyzer/cases.py:62
+          - src/anomalyzer/cases.py:115
+          - src/anomalyzer/cases.py:118
+          - src/anomalyzer/mcp_views.py:76
+          - src/anomalyzer/mcp_views.py:152
+          - src/anomalyzer/mcp_views.py:380
+          - src/anomalyzer/mcp_views.py:385
+          - tests/test_events.py:29
+          - tests/test_events.py:30
+          - tests/test_events.py:80
+          - tests/test_events.py:81
+          - tests/test_events.py:121
+          - docs/sparse-data.md:118
+          - docs/sparse-data.md:136
+          - docs/sparse-data.md:137
+          - docs/sparse-data.md:138
+          - docs/sparse-data.md:139
+          - examples/sparse_examples.py:87
+          - examples/sparse_examples.py:88
+          - examples/sparse_examples.py:89
+          - examples/render_readme_progression.py:918
+          - examples/render_readme_progression.py:921
+          - examples/render_readme_progression.py:1024
+          - examples/render_readme_progression.py:1037
+          - examples/render_readme_progression.py:1038
+        notes: Related implementation text was found, but deterministic matching cannot
+          prove the behavior. Review the cited evidence. If it is correct, save
+          agent.review evidence or make the criterion/reference wording more
+          explicit.
+      - criterion: Event-specific evidence and case identities survive deterministic
+          replay; availability and identified corrections retain as-of
+          provenance and bounded results.
+        status: partial
+        evidence:
+          - src/anomalyzer/contracts.py:308
+          - src/anomalyzer/contracts.py:315
+          - src/anomalyzer/contracts.py:319
+          - src/anomalyzer/contracts.py:596
+          - src/anomalyzer/contracts.py:597
+          - src/anomalyzer/events.py:1
+          - src/anomalyzer/events.py:14
+          - src/anomalyzer/events.py:100
+          - src/anomalyzer/events.py:102
+          - src/anomalyzer/events.py:165
+          - src/anomalyzer/sparse.py:182
+          - src/anomalyzer/sparse.py:194
+          - src/anomalyzer/sparse.py:195
+          - src/anomalyzer/sparse.py:202
+          - src/anomalyzer/sparse.py:333
+          - src/anomalyzer/orchestrator.py:122
+          - src/anomalyzer/orchestrator.py:143
+          - src/anomalyzer/orchestrator.py:155
+          - src/anomalyzer/orchestrator.py:162
+          - src/anomalyzer/orchestrator.py:164
+          - src/anomalyzer/alignment.py:63
+          - src/anomalyzer/alignment.py:66
+          - src/anomalyzer/relationships.py:35
+          - src/anomalyzer/relationships.py:198
+          - src/anomalyzer/relationships.py:203
+          - src/anomalyzer/relationships.py:205
+          - src/anomalyzer/cases.py:29
+          - src/anomalyzer/cases.py:30
+          - src/anomalyzer/cases.py:33
+          - src/anomalyzer/cases.py:37
+          - src/anomalyzer/cases.py:62
+          - src/anomalyzer/mcp_views.py:12
+          - src/anomalyzer/mcp_views.py:14
+          - src/anomalyzer/mcp_views.py:76
+          - src/anomalyzer/mcp_views.py:77
+          - src/anomalyzer/mcp_views.py:88
+          - tests/test_events.py:1
+          - tests/test_events.py:12
+          - tests/test_events.py:36
+          - tests/test_events.py:49
+          - tests/test_events.py:51
+          - docs/sparse-data.md:37
+          - docs/sparse-data.md:97
+          - docs/sparse-data.md:100
+          - docs/sparse-data.md:111
+          - docs/sparse-data.md:115
+          - examples/sparse_examples.py:39
+          - examples/sparse_examples.py:72
+          - examples/sparse_examples.py:73
+          - examples/sparse_examples.py:77
+          - examples/sparse_examples.py:80
+          - examples/render_readme_progression.py:3
+          - examples/render_readme_progression.py:781
+          - examples/render_readme_progression.py:782
+          - examples/render_readme_progression.py:787
+          - examples/render_readme_progression.py:798
+        notes: Related implementation text was found, but deterministic matching cannot
+          prove the behavior. Review the cited evidence. If it is correct, save
+          agent.review evidence or make the criterion/reference wording more
+          explicit.
+    gaps:
+      - Identified event-index sources retain every distinct event and its
+        original time, reject repeated logical IDs and preserve calendar and
+        legacy semantics. Review the cited evidence. If it is correct, save
+        agent.review evidence or make the criterion/reference wording more
+        explicit.
+      - Without a trusted sequence, whole timestamp cohorts share prior-time
+        point references and ties cannot establish change, Nelson or CUSUM
+        continuity. Review the cited evidence. If it is correct, save
+        agent.review evidence or make the criterion/reference wording more
+        explicit.
+      - Explicit trusted sequence metadata establishes a validated event order,
+        not independence, elapsed cadence or operational authority. Review the
+        cited evidence. If it is correct, save agent.review evidence or make the
+        criterion/reference wording more explicit.
+      - Event-specific evidence and case identities survive deterministic
+        replay; availability and identified corrections retain as-of provenance
+        and bounded results. Review the cited evidence. If it is correct, save
+        agent.review evidence or make the criterion/reference wording more
+        explicit.
diff --git a/.capabilities/data/quality/prepare-timestamped-series.capability.yaml b/.capabilities/data/quality/prepare-timestamped-series.capability.yaml
index 616b4e1..e14f430 100644
--- a/.capabilities/data/quality/prepare-timestamped-series.capability.yaml
+++ b/.capabilities/data/quality/prepare-timestamped-series.capability.yaml
@@ -9,8 +9,9 @@ acceptance:
     without a calendar frequency.
   - Local timestamps require a timezone, and ambiguous or nonexistent
     daylight-saving times require an explicit offset.
-  - Duplicate timestamps are rejected unless an aggregation policy is supplied;
-    duplicate counts and transformations remain visible.
+  - Calendar duplicate timestamps require an aggregation policy; identified
+    event-index calls can share original timestamps while duplicate logical IDs
+    are rejected and arbitrary tie order supplies no sequence evidence.
   - An as_of cutoff excludes later observations, and cadence is checked as a
     fixed elapsed duration.
 guidance:
@@ -18,6 +19,8 @@ guidance:
     Schema-1.1 source policies may explicitly authorize confirmed empty additive
     volume as zero.
   - Duplicate null values propagate through aggregation.
+  - Stable event identity and trusted sequence are separate caller contracts;
+    legacy event inputs without IDs still require unique timestamps.
 planning:
   story_map:
     backbone: Analyze a local series
@@ -37,6 +40,9 @@ agent:
   # -----------------------------
   implementation:
     references:
+      - src/anomalyzer/events.py
+      - tests/test_events.py
+      - docs/sparse-data.md
       - src/anomalyzer/sparse.py
       - tests/test_sparse.py
       - src/anomalyzer/validation.py
@@ -57,6 +63,7 @@ agent:
           cutoff.
         command: python -m pytest -q
           tests/test_analysis.py::test_duplicate_sort_and_cutoff
+          tests/test_events.py::test_distinct_same_time_calls_keep_values_references_and_cases
       - id: cadence-context
         description: Check fixed weekly cadence and preserve cutoff semantics across
           interfaces.
diff --git a/.capabilities/dependency-graph.svg b/.capabilities/dependency-graph.svg
index 1a36f80..9f3cc11 100644
--- a/.capabilities/dependency-graph.svg
+++ b/.capabilities/dependency-graph.svg
@@ -105,7 +105,7 @@
   </g>
   <text class="drag-note" x="52" y="950">Open this SVG directly in a browser for animation and dragging.</text>
   <script><![CDATA[
-const graph = {"nodes":[{"id":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","label":"Understand how much histor...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","acceptance":["Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","Provisional residual scores can be reported but do not trigger point-anomaly episodes.","Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger."],"guidance":["Evidence strength is not a probability of an incident.","Calibration and expectations are separate concerns; each declared regime calibrates again."],"path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/detection/identify-residual-patterns","analysis/detection/report-point-anomaly-episodes","analysis/evidence/assess-reference-readiness","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","relationships/comparisons/compare-declared-metrics"],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"short-history","description":"Keep reference-only evidence for a short series.","command":"python -m pytest -q tests/test_analysis.py::test_short_series_returns_reference_evidence"},{"id":"provisional-evidence","description":"Report provisional scores without point triggers.","command":"python -m pytest -q tests/test_analysis.py::test_early_evidence_maturity_and_no_provisional_trigger"},{"id":"independent-calibration","description":"Check a known residual center, sample scale, and standardized score.","command":"python -m pytest -q tests/test_analysis.py::test_calibration_reference"},{"id":"constant-reference","description":"Handle zero residual variance without spurious episodes.","command":"python -m pytest -q tests/test_analysis.py::test_constant_and_schema"}],"manualChecks":["Inspect a sample's signal_maturity, calibration_samples, standardized_residual, and triggers together before describing a statistical departure."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","criteria":[{"criterion":"Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","status":"covered","evidence":["src/anomalyzer/recipes.py:205","src/anomalyzer/recipes.py:445","tests/test_analysis.py:141"],"notes":"The short-series test checks three reference-only samples, absent standardized scores, and no triggers."},{"criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"covered","evidence":["src/anomalyzer/recipes.py:210","tests/test_analysis.py:123"],"notes":"A provisional score of 2 exceeds the configured point threshold of 1 but produces no trigger or episode."},{"criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"covered","evidence":["src/anomalyzer/recipes.py:231","src/anomalyzer/recipes.py:307","tests/test_analysis.py:111"],"notes":"Calibration uses retained residual mean and sample standard deviation once per regime; the independent numerical example produces the expected standardized score."},{"criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"covered","evidence":["src/anomalyzer/contracts.py:104","src/anomalyzer/recipes.py:350","src/anomalyzer/recipes.py:388","tests/test_analysis.py:111"],"notes":"Sample fields retain the comparison and support; the point threshold is inspectable in the enclosing method parameters."}],"ignore_findings":[],"done":true,"input_fingerprint":"ccdc1b2e92f3e647d5a9e6127cc3c4b10bf0da8f1f78af3efb805daa9b1e4305","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Establish evidence support"},"impact":6,"gaps":0,"r":68,"x":170,"y":170},{"id":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","label":"Bring my metrics into an a...","area":"data","scopes":["data","data/inputs"],"scope":"data/inputs","scopeLabel":"data/inputs","status":"implemented","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","acceptance":["Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","Published schema 1.0 and 1.1 request and result schemas match the public Python contracts."],"guidance":["Null observations remain unknown unless a schema-1.1 additive-volume policy explicitly authorizes confirmed empty collection as zero. Limits apply to the request as a whole.","Describe current supported behavior; historical plans do not add accepted inputs."],"path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["analysis/evidence/calibrate-departures","analysis/expectations/model-seasonal-level-and-growth","data/quality/analyze-sparse-observations","execution/resilience/retain-evidence-under-failure-and-limits","interfaces/python/analyze-through-public-api","investigation/assessments/describe-early-changes"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_agentic.py","schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json"],"automatedChecks":[{"id":"input-forms","description":"Analyze a bare list and validate a canonical request and result.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings tests/test_analysis.py::test_constant_and_schema"},{"id":"explicit-scope","description":"Reject schema 1.1 datasets without entity scope.","command":"python -m pytest -q tests/test_agentic.py::test_schema_11_requires_explicit_entity_scope"},{"id":"invalid-inputs","description":"Reject unsupported settings and nonnumeric or nonfinite observations.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_config tests/test_analysis.py::test_invalid_values"},{"id":"published-schemas","description":"Compare all four checked-in schemas with their public contracts.","command":"python -m pytest -q tests/test_analysis.py::test_checked_in_schemas tests/test_agentic.py::test_v11_checked_in_schemas_exist"}],"manualChecks":["Review the input examples in README.md and confirm they describe the two supported schema versions and the four-dataset/four-relationship limit."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","criteria":[{"criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"covered","evidence":["src/anomalyzer/recipes.py:21","src/anomalyzer/contracts.py:96","tests/test_analysis.py:234"],"notes":"Bare-list settings are validated; schema 1.0 limits datasets to exactly one."},{"criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"covered","evidence":["src/anomalyzer/contracts.py:240","src/anomalyzer/contracts.py:335","tests/test_agentic.py:192"],"notes":"Entity scope is required, relationship kinds are discriminated, and references must name distinct known datasets."},{"criterion":"Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:7","src/anomalyzer/contracts.py:71","tests/test_analysis.py:526","tests/test_analysis.py:531"],"notes":"Assertions reject each listed category; missing null values remain a separate accepted representation."},{"criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"covered","evidence":["schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json","tests/test_analysis.py:547","tests/test_agentic.py:317"],"notes":"Tests and independent inspection compare each complete published schema with its contract-generated schema. Python cross-field validators supplement JSON Schema constraints."}],"ignore_findings":[],"done":true,"input_fingerprint":"3546cd8f436812ff11c9352a855926677b55cc6287e40f6a5041bd613173e5b4","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Supply a numeric series"},"impact":6,"gaps":0,"r":68,"x":353.33333333333337,"y":170},{"id":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","label":"Analyze local files from t...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","acceptance":["The methods command lists supported analysis recipes and residual detectors.","Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status."],"guidance":["CSV callers must select one entity and unit before analysis.","Exit 0 includes completed analysis with anomalies, insufficient history, or inapplicability; exit 2 means invalid input, and exit 1 means execution failure or partial execution."],"path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/python/install-versioned-library","interfaces/skill/guide-anomaly-investigation","interfaces/skill/visualize-analysis-on-request"],"implementationReferences":["src/anomalyzer/cli.py","src/anomalyzer/io.py","src/anomalyzer/__main__.py","tests/test_cli.py"],"automatedChecks":[{"id":"file-stdin","description":"Analyze equivalent file and stdin requests.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence"},{"id":"csv-precedence","description":"Validate selected CSV fields and apply flag-over-file-over-embedded precedence.","command":"python -m pytest -q tests/test_cli.py::test_csv_and_precedence"},{"id":"value-list","description":"Analyze a bare JSON array and expose inferred settings.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"},{"id":"output-and-exits","description":"Protect existing output and distinguish invalid options from failed execution.","command":"python -m pytest -q tests/test_cli.py::test_output_protection tests/test_cli.py::test_methods_and_invalid_options tests/test_cli.py::test_size_limit_and_execution_exit"}],"manualChecks":["Run an example in readable and JSON modes and confirm stdout contains results while diagnostics go to stderr."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","criteria":[{"criterion":"The methods command lists supported analysis recipes and residual detectors.","status":"covered","evidence":["src/anomalyzer/cli.py:426","src/anomalyzer/registry.py:15","tests/test_cli.py:27"],"notes":"Discovery tests check all three recipes and ten residual detectors, with structured stdout and no successful-run diagnostics on stderr."},{"criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:239","tests/test_cli.py:13","tests/test_mcp.py:163"],"notes":"The passing file/stdin regression compares request fingerprints and short-history semantics; cross-interface tests compare complete stable numerical output."},{"criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"covered","evidence":["src/anomalyzer/io.py:33","tests/test_cli.py:51"],"notes":"The passing CSV regression covers timestamped missing cells, positional data, and varying unmapped entity rejection."},{"criterion":"Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","status":"covered","evidence":["src/anomalyzer/cli.py:253","tests/test_cli.py:51","tests/test_cli.py:71"],"notes":"The passing file/flag/embedded precedence regression and bare-values regression establish the merge order and validated configuration."},{"criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"covered","evidence":["src/anomalyzer/io.py:69","src/anomalyzer/cli.py:424","tests/test_cli.py:41","tests/test_cli.py:108"],"notes":"Output protection and exit regressions passed, including explicit overwrite; successful manual examples have empty stderr and incomplete-usage exits 1."}],"ignore_findings":[],"done":true,"input_fingerprint":"86e701ccc0a1aefd407f11b4547784fef497fa1b49ddf6bb69db0d08f757dfdc","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Run from files or stdin"},"impact":4,"gaps":0,"r":58,"x":903.3333333333334,"y":170},{"id":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","label":"Account for ordinary seaso...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","acceptance":["A supplied season length selects earlier matching seasonal observations for the reference.","Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected."],"guidance":["Expectations support scoring; they are not a standalone forecasting API."],"path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/expectations/protect-reference-from-extremes","analysis/expectations/reset-declared-regimes","analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py"],"automatedChecks":[{"id":"seasonal-reference","description":"Check independent lagged-reference values.","command":"python -m pytest -q tests/test_analysis.py::test_forecast_independent_reference"},{"id":"growth-and-departures","description":"Keep clean linear and compound growth ordinary while identifying injected departures.","command":"python -m pytest -q tests/test_trend.py::test_trend_and_anomalies"},{"id":"frozen-fit","description":"Keep the training fit stable when later observations change.","command":"python -m pytest -q tests/test_trend.py::test_frozen_fit_and_legacy_mode"},{"id":"explicit-trends","description":"Accept supported trend choices and handle short, constant, and signed series.","command":"python -m pytest -q tests/test_trend.py::test_explicit_modes_and_validation tests/test_trend.py::test_short_history_constant_and_signed_linear"}],"manualChecks":["Compare examples/growth_up.json and examples/compound.json with their selected trend and evidence expectations."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","criteria":[{"criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"covered","evidence":["src/anomalyzer/baseline.py:8","src/anomalyzer/baseline.py:31","tests/test_analysis.py:21"],"notes":"Independent lag values and robust matching-phase assertions establish the seasonal reference calculation."},{"criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"covered","evidence":["src/anomalyzer/trend.py:120","src/anomalyzer/trend.py:144","tests/test_trend.py:11"],"notes":"Positive and negative linear/exponential cases select the expected model, match post-training expectations, and produce no ordinary-growth episodes."},{"criterion":"The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","status":"covered","evidence":["src/anomalyzer/recipes.py:162","tests/test_trend.py:36"],"notes":"Trend fitting occurs at the training boundary; the test asserts identical trend diagnostics and earlier evidence after later observations change."},{"criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:20","tests/test_trend.py:49","tests/test_trend.py:57"],"notes":"Tests exercise disabled, linear, exponential, and invalid quadratic modes, including disclosed short-history fallback."}],"ignore_findings":[],"done":true,"input_fingerprint":"8e9a0361b02b7d3f651366e5fa0675cccb3949e44de448d2ae0d43e6710bd408","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Account for growth and seasonality"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","label":"Embed local analysis in my...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","acceptance":["analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","analyze_series preserves supplied settings and event-time context for a single dataset.","analyze_relationships preserves the declared multi-dataset relationship semantics.","Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime."],"guidance":["All interfaces use local analysis; there is no hosted service or model API requirement."],"path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence","interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/__init__.py","src/anomalyzer/agent.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"request-parity","description":"Match canonical request results across Python, CLI, and full-result MCP compatibility calls.","command":"python -m pytest -q tests/test_mcp.py::test_canonical_request_matches_python_cli_and_mcp tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"series-context","description":"Preserve a single-series event-time cutoff and context across interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp tests/test_mcp.py::test_series_wrapper_preserves_omitted_and_explicit_season_settings"},{"id":"relationship-parity","description":"Match declared relationship evidence across the public interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_relationship_semantics_match_python_cli_and_mcp"}],"manualChecks":["Review the Python API examples in README.md and confirm callers can inspect the returned validated models and JSON."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","criteria":[{"criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"covered","evidence":["src/anomalyzer/__init__.py:3","src/anomalyzer/recipes.py:21","tests/test_analysis.py:234"],"notes":"Bare values and separate settings validate before dispatch; canonical schema 1.0 and 1.1 requests use their respective contracts."},{"criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"covered","evidence":["src/anomalyzer/agent.py:11","tests/test_mcp.py:191","tests/test_mcp.py:291"],"notes":"Field-set handling preserves omitted versus explicit settings and context in the single-series wrapper, verified across dictionary and Settings inputs."},{"criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"covered","evidence":["src/anomalyzer/agent.py:27","tests/test_mcp.py:217"],"notes":"Declared relationships validate and use the shared engine without changing units, lineage, or primitive semantics."},{"criterion":"Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime.","status":"covered","evidence":["tests/test_mcp.py:163","tests/test_mcp.py:191","tests/test_mcp.py:217","src/anomalyzer/__main__.py:1"],"notes":"Passing real CLI subprocess, Python, and full-result MCP parity tests compare stable numerical results apart from identity and runtime."}],"ignore_findings":[],"done":true,"input_fingerprint":"214bd10bb3aa5fad0f74938fe2ef09853a593cdb5b93cba49ec851a628d8ab65","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","name":"Python integration","label":"Python integration / Embed analysis in my application / Call the public API"},"impact":3,"gaps":0,"r":53,"x":903.3333333333334,"y":170},{"id":"investigation/cases/group-related-evidence","title":"Investigate related findings together","label":"Investigate related findin...","area":"investigation","scopes":["investigation","investigation/cases"],"scope":"investigation/cases","scopeLabel":"investigation/cases","status":"implemented","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","acceptance":["Evidence for different declared entities remains in separate cases even at the same event time.","Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","A derived relationship adds context without counting as an additional independent source for severity.","Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection."],"guidance":["Warning and critical are deterministic numerical grouping rules, not validated incident impact.","Diagnostic trend, oscillation, mixture, and variation findings remain distinct from adaptive location-shift cases."],"path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","dependencies":["investigation/assessments/describe-early-changes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/mcp/investigate-findings-and-evidence","investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/cases.py","src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py"],"automatedChecks":[{"id":"entity-scope","description":"Separate cases for different entities at the same event time.","command":"python -m pytest -q tests/test_agentic.py::test_cases_are_scoped_by_entity_at_the_same_event_time"},{"id":"source-severity","description":"Escalate numerical severity when two source datasets meet criteria.","command":"python -m pytest -q tests/test_agentic.py::test_two_source_datasets_escalate_case_severity"},{"id":"correlated-context","description":"Keep relationship evidence from double-counting source severity.","command":"python -m pytest -q tests/test_agentic.py::test_relationship_evidence_explains_but_does_not_double_count_severity"},{"id":"first-detection","description":"Group an adaptive location shift before point triggers without using a later pattern peak.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_location_shift_creates_one_early_case_without_point_trigger tests/test_agentic.py::test_location_case_severity_does_not_use_future_pattern_peak"}],"manualChecks":["Inspect a case's contributing targets, explanation, evidence references, and investigation questions; distinguish it from a point-anomaly episode."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","criteria":[{"criterion":"Evidence for different declared entities remains in separate cases even at the same event time.","status":"covered","evidence":["src/anomalyzer/cases.py:42","tests/test_agentic.py:69"],"notes":"Entity participates in grouping and stable identity; the test requires separate cases and distinct IDs at the same time."},{"criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"covered","evidence":["src/anomalyzer/cases.py:72","tests/test_agentic.py:85"],"notes":"Two qualifying source datasets produce critical numerical severity and named source explanations."},{"criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"covered","evidence":["src/anomalyzer/cases.py:72","src/anomalyzer/cases.py:116","tests/test_agentic.py:100"],"notes":"The independent-source count includes dataset criteria only. Derived evidence remains explanatory and can still meet the separately declared magnitude rule."},{"criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"covered","evidence":["src/anomalyzer/early.py:193","src/anomalyzer/early.py:237","src/anomalyzer/orchestrator.py:115","tests/test_agentic.py:122","tests/test_agentic.py:146"],"notes":"Tests require one moderate-shift case without point triggers and restrict assessment references/strength to samples at or before first detection."}],"ignore_findings":[],"done":true,"input_fingerprint":"6830af8a2f064123c12adb062f4793e446b97b3b805fd73a934ea84cd0d5d9cf","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Investigate related evidence"},"impact":3,"gaps":0,"r":53,"x":1086.6666666666667,"y":170},{"id":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","label":"Spot sustained shifts and ...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","acceptance":["Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","Trend, oscillation, and mixture findings retain their specific pattern meanings.","Moving-range findings identify variation changes separately from location changes."],"guidance":["Pattern findings may overlap and are not independent confirmations.","A residual location shift does not establish a change in the full distribution."],"path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations","investigation/cases/group-related-evidence"],"implementationReferences":["src/anomalyzer/evidence.py","src/anomalyzer/recipes.py","tests/test_analysis.py"],"automatedChecks":[{"id":"location-shift","description":"Find moderate increases and decreases without point anomalies.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_patterns_describe_location_shift_without_point_anomaly"},{"id":"accumulating-change","description":"Detect a known CUSUM accumulation at the expected position.","command":"python -m pytest -q tests/test_analysis.py::test_cusum_detects_accumulated_moderate_location_shift"},{"id":"diagnostic-patterns","description":"Distinguish residual trend, oscillation, and mixture rules.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_diagnostics_have_specific_semantics"},{"id":"variation","description":"Describe a moving-range increase as variation rather than location.","command":"python -m pytest -q tests/test_analysis.py::test_moving_range_detects_short_term_variation_increase"}],"manualChecks":["Compare examples/location_shift_up.json with its point episodes and pattern findings, allowing for overlapping residual rules."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","criteria":[{"criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"covered","evidence":["src/anomalyzer/evidence.py:118","src/anomalyzer/evidence.py:189","tests/test_analysis.py:293"],"notes":"Tests assert no point triggers or episodes and exact upward/downward detection positions for Nelson rules and CUSUM."},{"criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"covered","evidence":["src/anomalyzer/evidence.py:211","src/anomalyzer/evidence.py:230","tests/test_analysis.py:350"],"notes":"The known accumulation detects at index 16 and reports peak CUSUM approximately 6, threshold 5, and increase direction."},{"criterion":"Trend, oscillation, and mixture findings retain their specific pattern meanings.","status":"covered","evidence":["src/anomalyzer/evidence.py:70","src/anomalyzer/evidence.py:122","tests/test_analysis.py:340"],"notes":"Separate predicates and descriptions preserve the three meanings and explicitly limit location/distribution claims."},{"criterion":"Moving-range findings identify variation changes separately from location changes.","status":"covered","evidence":["src/anomalyzer/evidence.py:234","tests/test_analysis.py:360"],"notes":"Adjacent standardized-residual differences produce variation_shift findings; the test checks magnitude, detection index, and interpretation."}],"ignore_findings":[],"done":true,"input_fingerprint":"63716eb67db6b73964b79897d8c81f2891df2586ff213c95ea81c5699b8abbea","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Find sustained residual changes"},"impact":2,"gaps":0,"r":48,"x":170,"y":170},{"id":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","label":"Analyze observations in th...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","acceptance":["Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration."],"guidance":["No interpolation or implicit filling, resampling, or unit conversion occurs. Schema-1.1 source policies may explicitly authorize confirmed empty additive volume as zero.","Duplicate null values propagate through aggregation."],"path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["data/quality/exclude-incomplete-periods","relationships/quality/require-compatible-sources"],"implementationReferences":["src/anomalyzer/sparse.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/contracts.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"coordinate-contract","description":"Reject partial timestamp metadata and positional as_of cutoffs.","command":"python -m pytest -q tests/test_analysis.py::test_timestamp_metadata_is_all_or_nothing"},{"id":"local-time","description":"Normalize valid local times and reject ambiguous or nonexistent times.","command":"python -m pytest -q tests/test_analysis.py::test_local_time"},{"id":"duplicates-cutoff","description":"Require an explicit duplicate policy and retain counts after the cutoff.","command":"python -m pytest -q tests/test_analysis.py::test_duplicate_sort_and_cutoff"},{"id":"cadence-context","description":"Check fixed weekly cadence and preserve cutoff semantics across interfaces.","command":"python -m pytest -q tests/test_analysis.py::test_weekly_cadence tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp"}],"manualChecks":["Inspect a timestamped result's data_quality transformations, UTC window, and resolved context to confirm preparation is explainable."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","criteria":[{"criterion":"Timestamped datasets provide both timestamps and frequency; positional datasets provide neither.","status":"covered","evidence":["src/anomalyzer/contracts.py:77","tests/test_analysis.py:418"],"notes":"Partial coordinate metadata and positional as_of are rejected."},{"criterion":"Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","status":"covered","evidence":["src/anomalyzer/validation.py:8","tests/test_analysis.py:500"],"notes":"UTC round trips detect ambiguous/nonexistent local times; valid local time normalizes to the expected UTC hour."},{"criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"covered","evidence":["src/anomalyzer/validation.py:87","src/anomalyzer/validation.py:103","tests/test_analysis.py:478"],"notes":"Explicit mean/sum handling retains duplicate counts and transformation descriptions. Nulls propagate rather than becoming zero."},{"criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"covered","evidence":["src/anomalyzer/validation.py:105","src/anomalyzer/validation.py:111","tests/test_analysis.py:491","tests/test_mcp.py:191"],"notes":"Filtering is inclusive through as_of; cadence compares UTC elapsed timedeltas, including fixed seven-day weeks."}],"ignore_findings":[],"done":true,"input_fingerprint":"791748764cdc0dfe81d987a908179930d858670606e51b64407172295776b261","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Prepare the comparison"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","label":"Let an agent inspect the e...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","acceptance":["The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","get_evidence exposes the selected finding's numerical support and relationship source lineage.","Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated."],"guidance":["Run results are retained only in the bounded server-session cache; missing or evicted run IDs require reanalysis.","Cases, episodes, and residual patterns may overlap."],"path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":["interfaces/mcp/connect-read-only-local-server","interfaces/skill/guide-anomaly-investigation"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py"],"automatedChecks":[{"id":"case-drilldown","description":"Analyze, page findings, and inspect exact conversion case values and lineage.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case tests/test_mcp.py::test_schema_11_episode_is_paged_and_has_relationship_evidence"},{"id":"unavailable-finding","description":"Expose a quality finding when latest relationship evidence is unavailable.","command":"python -m pytest -q tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"},{"id":"early-candidate","description":"Inspect descriptive candidate evidence without inventing a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case tests/test_mcp.py::test_sparse_insufficient_history_keeps_descriptive_candidate_visible"},{"id":"sample-pagination","description":"Page supporting samples for a residual pattern.","command":"python -m pytest -q tests/test_mcp.py::test_get_evidence_pages_pattern_samples"}],"manualChecks":["Follow analyze, list_findings, and get_evidence for the conversion example and check that the headline is supported by the retrieved values."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","criteria":[{"criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"covered","evidence":["src/anomalyzer/mcp.py:170","src/anomalyzer/mcp_views.py:247","tests/test_mcp.py:54"],"notes":"The conversion manual call returns a relationship headline, preview, counts, readiness, and session-scoped run ID."},{"criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:132","src/anomalyzer/mcp_views.py:260","tests/test_mcp.py:313","tests/test_mcp.py:108"],"notes":"Cases, early candidates, episodes, patterns, and quality findings feed a bounded page; relationship episode regression verifies schema 1.1 pagination."},{"criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:273","src/anomalyzer/mcp_views.py:294","tests/test_mcp.py:54","tests/test_mcp.py:313"],"notes":"Manual drilldown returns observed 8/130 versus expected 0.1 and source indexes orders=14 and qualified_visits=14."},{"criterion":"Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:118","src/anomalyzer/mcp_views.py:169","src/anomalyzer/mcp_views.py:337","tests/test_mcp.py:87","tests/test_mcp.py:137","tests/test_mcp.py:341"],"notes":"Passing tests retain short-history and inapplicable-method candidates plus unavailable latest evidence, while pattern samples are paginated."}],"ignore_findings":[],"done":true,"input_fingerprint":"a09b0003780bcd2f9ccf8e3d8ff58b3147c118760b984757d7027b5810fd9bef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Drill into findings and evidence"},"impact":2,"gaps":0,"r":48,"x":927.3333333333334,"y":288},{"id":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","label":"Spot early changes with a ...","area":"investigation","scopes":["investigation","investigation/assessments"],"scope":"investigation/assessments","scopeLabel":"investigation/assessments","status":"implemented","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","acceptance":["The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","Missing values break the comparison history, and zero previous values do not create relative-change evidence.","Early candidates remain non-triggering and ineligible for action even under a permissive policy."],"guidance":["An early candidate does not establish an anomaly, stable baseline, cause, or incident."],"path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["investigation/cases/group-related-evidence","investigation/rules/check-declared-numerical-rules"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_mcp.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"prefix-invariant-facts","description":"Describe short history causally without creating cases.","command":"python -m pytest -q tests/test_agentic.py::test_short_history_assessments_are_useful_and_prefix_invariant"},{"id":"comparison-boundaries","description":"Reset comparisons at missing observations and avoid dividing by zero.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"candidate-eligibility","description":"Keep candidates ineligible for action and notifications under permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"},{"id":"agent-visibility","description":"Expose a short-history candidate and its claim limits without a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case"}],"manualChecks":["Inspect a candidate's comparison population, maturity, established claim, and not_established text before describing it."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","criteria":[{"criterion":"The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","status":"covered","evidence":["src/anomalyzer/early.py:73","src/anomalyzer/early.py:96","tests/test_agentic.py:35"],"notes":"Assertions establish observation/change/change/departure_candidate for [10,12,11,20]."},{"criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"covered","evidence":["src/anomalyzer/early.py:50","src/anomalyzer/early.py:141","tests/test_agentic.py:35"],"notes":"Each assessment precedes the prior-history update; tests compare complete serialized assessments for every prefix."},{"criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"covered","evidence":["src/anomalyzer/early.py:53","src/anomalyzer/early.py:58","tests/test_capability_behaviors.py:87"],"notes":"The zero comparison has no relative change; after a missing value, comparison samples reset to zero."},{"criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"covered","evidence":["src/anomalyzer/early.py:105","src/anomalyzer/cases.py:35","tests/test_capability_behaviors.py:134","tests/test_mcp.py:341"],"notes":"Candidates have criterion_met=false and remain action/notification-ineligible after policy replay, without creating cases."}],"ignore_findings":[],"done":true,"input_fingerprint":"debdeeeaea559950f53461a4298cb9c2e3a1681f83a43e8a981ed9b51968a4d0","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect early evidence"},"impact":2,"gaps":0,"r":48,"x":1110.6666666666667,"y":288},{"id":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","label":"Compare the same entity an...","area":"relationships","scopes":["relationships","relationships/quality"],"scope":"relationships/quality","scopeLabel":"relationships/quality","status":"implemented","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","acceptance":["Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","Difference relationships require identical declared source units and matching output units.","An invalid relationship is reported as failed while valid sibling dataset results remain available."],"guidance":["A failed or unavailable relationship does not establish that its metric is normal.","No implicit timestamp matching, interpolation, unit conversion, or zero-filling occurs."],"path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"implementationReferences":["src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"exact-alignment","description":"Reject mismatched timestamps, cadence, entity, duplicates, and positional relationship sources.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relationship_requires_exact_source_alignment"},{"id":"compatible-units","description":"Report a difference unit mismatch without suppressing sibling results.","command":"python -m pytest -q tests/test_agentic.py::test_alignment_unit_and_partial_failure_are_specific"},{"id":"unambiguous-declaration","description":"Reject fields that do not belong to the selected relationship kind.","command":"python -m pytest -q tests/test_agentic.py::test_contract_rejects_ambiguous_relationship_fields"},{"id":"healthy-siblings","description":"Retain a healthy dataset's evidence when another dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Read both target-level statuses and errors before interpreting a partial run; check source dataset errors when a relationship source failed."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","criteria":[{"criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"covered","evidence":["src/anomalyzer/alignment.py:42","tests/test_capability_behaviors.py:30"],"notes":"Parameterized assertions require specific errors for timestamp, cadence, and entity mismatches."},{"criterion":"Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","status":"covered","evidence":["src/anomalyzer/alignment.py:23","src/anomalyzer/alignment.py:44","tests/test_capability_behaviors.py:30"],"notes":"Positional sources fail exact alignment; even an explicit duplicate-mean setting is rejected for relationship source lineage."},{"criterion":"Difference relationships require identical declared source units and matching output units.","status":"covered","evidence":["src/anomalyzer/relationships.py:41","tests/test_agentic.py:242"],"notes":"Source and output unit checks are explicit. Tests cover mismatched source units and successful matching units."},{"criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:171","tests/test_capability_behaviors.py:30","tests/test_capability_behaviors.py:159"],"notes":"Relationship errors remain target-specific, healthy dataset results survive, and the overall mixed run is partial."}],"ignore_findings":[],"done":true,"input_fingerprint":"06321ab5806a8276cc3954df5d6f4dd7f059f67600038854569787ff51b420d6","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Align comparable sources"},"impact":2,"gaps":0,"r":48,"x":1294,"y":288},{"id":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","label":"Compare against a new base...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","acceptance":["Integer positions and exact observed dates or timestamps can identify reset boundaries.","Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","Repeated, out-of-range, or unmatched reset points are rejected.","CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries."],"guidance":["Reset points are explicit caller decisions. Known-event notes and review-state fields do not rebuild regimes.","Multi-resolution analysis does not support reset_points."],"path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations"],"implementationReferences":["src/anomalyzer/validation.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py"],"automatedChecks":[{"id":"independent-regimes","description":"Retrain separate regimes without cross-boundary evidence or patterns.","command":"python -m pytest -q tests/test_analysis.py::test_manual_resets_retrain_without_crossing_boundaries"},{"id":"invalid-boundaries","description":"Reject invalid or duplicate resolved reset positions.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_manual_resets"},{"id":"cli-boundaries","description":"Resolve position and date reset flags through the CLI.","command":"python -m pytest -q tests/test_cli.py::test_manual_reset_cli"}],"manualChecks":["Inspect the resolved reset positions and segment diagnostics for examples/steps_reset.json."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","criteria":[{"criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"covered","evidence":["src/anomalyzer/validation.py:129","tests/test_analysis.py:381"],"notes":"Positions and observed dates are exercised end to end; timestamp normalization and exact matching are implemented directly."},{"criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"covered","evidence":["src/anomalyzer/recipes.py:119","src/anomalyzer/recipes.py:143","tests/test_analysis.py:381"],"notes":"Each segment initializes separate state and passes its own start to forecasts. Tests verify three calibrated regimes and no cross-boundary evidence or findings."},{"criterion":"Repeated, out-of-range, or unmatched reset points are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/validation.py:149","tests/test_analysis.py:413"],"notes":"Validation rejects literal duplicates and different declarations resolving to the same position, invalid positions, and unmatched dates/timestamps."},{"criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"covered","evidence":["tests/test_cli.py:82","src/anomalyzer/recipes.py:53"],"notes":"CLI assertions establish positions [50, 100] and removal of the unmarked-regime episodes."}],"ignore_findings":[],"done":true,"input_fingerprint":"067063f972c5b70d990d6cb0125892731a97a7f482698d7b331a1a30229ac223","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Restart after a reviewed change"},"impact":1,"gaps":0,"r":43,"x":194,"y":288},{"id":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","label":"Assess sparse observations...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","acceptance":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"guidance":["Default missingness is unknown. A modeled expectation is never a measurement.","Caller business severity, incident state and notifications remain outside the engine."],"path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","src/anomalyzer/orchestrator.py","src/anomalyzer/relationships.py","src/anomalyzer/early.py","tests/test_sparse.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"source-semantics","description":"Preserve unknown periods and require explicit empty-volume authorization.","command":"python -m pytest -q tests/test_sparse.py -k \"zero or omitted or grid or duplicate or expansion\""},{"id":"causal-boundaries","description":"Retain real phase, event times, resets and as-of revision eligibility.","command":"python -m pytest -q tests/test_sparse.py -k \"phase or gap or reset or event or revision or arrival\""},{"id":"public-interfaces","description":"Preserve useful rules, relationships, views, cases and API parity.","command":"python -m pytest -q tests/test_sparse.py -k \"rule or relationship or daily or parity\""},{"id":"progressive-examples","description":"Reproduce the full numbered sequence and sparse worked examples.","command":"python examples/run_demo.py"}],"manualChecks":["Inspect progression figures 14–22 for true gaps, distinguished policy zeros, original event time, unavailable checks and caller-owned business severity."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/analyze-sparse-observations.","criteria":[{"criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","status":"partial","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:71","src/anomalyzer/contracts.py:85","src/anomalyzer/contracts.py:86","src/anomalyzer/sparse.py:3","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:42","src/anomalyzer/sparse.py:48","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse_views.py:74","src/anomalyzer/sparse_views.py:81","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:97","src/anomalyzer/sparse_views.py:119","src/anomalyzer/orchestrator.py:63","src/anomalyzer/orchestrator.py:127","src/anomalyzer/orchestrator.py:139","src/anomalyzer/orchestrator.py:191","src/anomalyzer/relationships.py:36","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:98","src/anomalyzer/relationships.py:131","src/anomalyzer/relationships.py:141","src/anomalyzer/early.py:58","src/anomalyzer/early.py:137","tests/test_sparse.py:44","tests/test_sparse.py:45","tests/test_sparse.py:50","tests/test_sparse.py:51","tests/test_sparse.py:57","docs/sparse-data.md:4","docs/sparse-data.md:29","docs/sparse-data.md:37","docs/sparse-data.md:39","docs/sparse-data.md:40","examples/sparse_examples.py:22","examples/sparse_examples.py:23","examples/sparse_examples.py:25","examples/sparse_examples.py:26","examples/sparse_examples.py:28","examples/render_readme_progression.py:135","examples/render_readme_progression.py:139","examples/render_readme_progression.py:140","examples/render_readme_progression.py:166","examples/render_readme_progression.py:206"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","status":"partial","evidence":["src/anomalyzer/contracts.py:27","src/anomalyzer/contracts.py:49","src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:56","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:181","src/anomalyzer/sparse.py:187","src/anomalyzer/sparse.py:189","src/anomalyzer/sparse_views.py:15","src/anomalyzer/sparse_views.py:98","src/anomalyzer/sparse_views.py:107","src/anomalyzer/sparse_views.py:108","src/anomalyzer/sparse_views.py:112","src/anomalyzer/orchestrator.py:22","src/anomalyzer/orchestrator.py:23","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:129","src/anomalyzer/orchestrator.py:141","src/anomalyzer/relationships.py:185","src/anomalyzer/relationships.py:189","src/anomalyzer/relationships.py:191","src/anomalyzer/early.py:46","src/anomalyzer/early.py:48","src/anomalyzer/early.py:51","src/anomalyzer/early.py:59","src/anomalyzer/early.py:61","tests/test_sparse.py:20","tests/test_sparse.py:108","tests/test_sparse.py:111","tests/test_sparse.py:113","tests/test_sparse.py:119","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:42","docs/sparse-data.md:106","docs/sparse-data.md:108","examples/sparse_examples.py:16","examples/sparse_examples.py:33","examples/sparse_examples.py:50","examples/sparse_examples.py:51","examples/sparse_examples.py:91","examples/render_readme_progression.py:241","examples/render_readme_progression.py:312","examples/render_readme_progression.py:317","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:242","src/anomalyzer/contracts.py:245","src/anomalyzer/contracts.py:258","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:57","src/anomalyzer/sparse.py:63","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:127","src/anomalyzer/sparse_views.py:155","src/anomalyzer/sparse_views.py:157","src/anomalyzer/orchestrator.py:37","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:178","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:40","src/anomalyzer/relationships.py:57","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:96","src/anomalyzer/relationships.py:97","src/anomalyzer/early.py:140","src/anomalyzer/early.py:147","tests/test_sparse.py:59","tests/test_sparse.py:67","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:128","docs/sparse-data.md:26","docs/sparse-data.md:96","docs/sparse-data.md:101","docs/sparse-data.md:132","docs/sparse-data.md:142","examples/sparse_examples.py:26","examples/sparse_examples.py:36","examples/sparse_examples.py:38","examples/sparse_examples.py:57","examples/sparse_examples.py:81","examples/render_readme_progression.py:654","examples/render_readme_progression.py:697","examples/render_readme_progression.py:884","examples/render_readme_progression.py:896","examples/render_readme_progression.py:917"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","status":"partial","evidence":["src/anomalyzer/contracts.py:13","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:297","src/anomalyzer/contracts.py:298","src/anomalyzer/contracts.py:437","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse.py:66","src/anomalyzer/sparse.py:83","src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:28","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:94","src/anomalyzer/orchestrator.py:106","src/anomalyzer/orchestrator.py:107","src/anomalyzer/orchestrator.py:133","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:160","src/anomalyzer/relationships.py:162","src/anomalyzer/relationships.py:166","tests/test_sparse.py:59","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:123","tests/test_sparse.py:180","docs/sparse-data.md:26","docs/sparse-data.md:41","docs/sparse-data.md:69","docs/sparse-data.md:78","docs/sparse-data.md:96","examples/sparse_examples.py:59","examples/sparse_examples.py:95","examples/render_readme_progression.py:686","examples/render_readme_progression.py:699","examples/render_readme_progression.py:710","examples/render_readme_progression.py:767","examples/render_readme_progression.py:884"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","partial: Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","partial: Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","partial: As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Assess sparse source data"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","label":"Avoid mistaking unfinished...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","acceptance":["Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","The observed values and timestamps of unfinished periods remain available in data quality output.","Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","A relationship with an unavailable latest source reports that unavailability rather than normal behavior."],"guidance":["Completeness is declared by the caller; the engine does not infer collection completeness."],"path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/alignment.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"unfinished-values","description":"Retain an unfinished value without scoring it.","command":"python -m pytest -q tests/test_analysis.py::test_explicit_incomplete_period_is_reported_and_not_scored"},{"id":"trailing-only","description":"Reject incomplete observations followed by completed observations.","command":"python -m pytest -q tests/test_analysis.py::test_incomplete_periods_must_be_trailing"},{"id":"source-and-agent-views","description":"Exclude a partial source sample from both analysis views and expose unavailable relationship evidence.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"}],"manualChecks":["Inspect examples/agentic/incomplete-usage.json and confirm the latest headline explains missing completeness and does not imply normality."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicitly incomplete trailing observations are reported as data quality evidence and excluded from completed-series analysis.","criteria":[{"criterion":"Incomplete observations form a trailing suffix after chronological preparation.","status":"covered","evidence":["src/anomalyzer/validation.py:44","tests/test_analysis.py:162"],"notes":"Suffix validation runs after timestamp preparation and rejects a completed observation following an incomplete one."},{"criterion":"The observed values and timestamps of unfinished periods remain available in data quality output.","status":"covered","evidence":["src/anomalyzer/validation.py:50","src/anomalyzer/validation.py:114","tests/test_analysis.py:150"],"notes":"Quality records retain original source index, normalized timestamp, observed value, and incomplete status."},{"criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"covered","evidence":["src/anomalyzer/validation.py:53","src/anomalyzer/recipes.py:564","tests/test_analysis.py:150","tests/test_analysis.py:200"],"notes":"Only finalized prepared values enter either scoring path; assertions verify the last scored sample and both multi-resolution counts."},{"criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"covered","evidence":["src/anomalyzer/alignment.py:35","src/anomalyzer/orchestrator.py:150","tests/test_mcp.py:87"],"notes":"Incomplete aligned sources become unavailable values, and latest relationship state/status plus MCP headline explain the lack of evidence."}],"ignore_findings":[],"done":true,"input_fingerprint":"d82ee816fe89a25232e2589587d31bcadfa81be4188a0e1531f4fbfdf0ebcb5e","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Keep unfinished data visible"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","label":"Give an agent read-only ac...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","acceptance":["Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","Analysis tool calls return structured content together with compact text.","Invalid tool input and unknown session run IDs return structured errors rather than fabricated results."],"guidance":["Compatibility tool names remain callable but are not advertised.","Legacy clients receive isError tool results; modern clients receive JSON-RPC invalid-input errors.","The server provides analysis and evidence inspection; the caller owns monitoring and operational workflows."],"path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","dependencies":["interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":["interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py","pyproject.toml"],"automatedChecks":[{"id":"tool-discovery","description":"Advertise the three read-only tools and their schema contracts.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_lists_read_only_tools"},{"id":"connection-protocols","description":"Support legacy stdio initialization and modern stateless discovery.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_stdio_legacy_handshake_and_tool_listing tests/test_mcp.py::test_mcp_modern_discovery_and_stateless_tool_listing"},{"id":"structured-content","description":"Return a compact analysis response with structured content and run ID.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_tool_call_returns_structured_content"},{"id":"invalid-calls","description":"Reject malformed input and unknown session run IDs in both protocols.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_invalid_tool_input_is_a_tool_error tests/test_mcp.py::test_paged_tool_inputs_enforce_schema_in_both_protocols"}],"manualChecks":["Review discovery's interpretation guidance and confirm the client can start anomalyzer-mcp in the installed Python environment."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","criteria":[{"criterion":"Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","status":"covered","evidence":["src/anomalyzer/mcp.py:63","tests/test_mcp.py:27"],"notes":"Discovery exposes three read-only tools with the same strict argument contracts used at dispatch and advertised output schemas."},{"criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"covered","evidence":["src/anomalyzer/mcp.py:228","tests/test_mcp.py:248","tests/test_mcp.py:268"],"notes":"Legacy initialization and modern stateless discovery tests establish protocol metadata and evidence interpretation guidance."},{"criterion":"Analysis tool calls return structured content together with compact text.","status":"covered","evidence":["src/anomalyzer/mcp.py:162","tests/test_mcp.py:39"],"notes":"The manual conversion call returns a compact headline and structured run ID, counts, readiness, and preview."},{"criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"covered","evidence":["src/anomalyzer/mcp.py:156","src/anomalyzer/mcp.py:257","tests/test_mcp.py:235","tests/test_mcp.py:367"],"notes":"Tests reject invalid cursor/limit types, unknown fields, numeric run IDs, and missing session runs in both protocols without fabricated results."}],"ignore_findings":[],"done":true,"input_fingerprint":"7bd92d238f4f9545372c6c35e78aefea60c585bea0a4a8bb99096ab6cef27f96","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Connect a local server"},"impact":1,"gaps":0,"r":43,"x":903.3333333333334,"y":406},{"id":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","label":"Find values that break my ...","area":"investigation","scopes":["investigation","investigation/rules"],"scope":"investigation/rules","scopeLabel":"investigation/rules","status":"implemented","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","acceptance":["Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","Absolute and relative changes must exceed their declared maximum to violate a rule.","Relative-change checks require a nonzero previous value and do not bridge missing observations.","Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them."],"guidance":["An exact rule crossing does not establish cause, an incident, or business impact."],"path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","dependencies":["investigation/assessments/describe-early-changes"],"suggestedDependencies":[],"dependents":["investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"rule-boundaries","description":"Check inclusive range boundaries and strict absolute/relative change thresholds.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_explicit_rules_respect_boundaries"},{"id":"relative-availability","description":"Avoid relative violations across zero or missing values.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"explicit-not-actionable","description":"Create evidence and a case for an exact rule crossing without enabling action by default.","command":"python -m pytest -q tests/test_agentic.py::test_explicit_rule_is_exact_but_not_actionable_without_policy"},{"id":"readable-rule","description":"Identify the exact unit-cost range rule in CLI output.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the unit-cost example's crossed rule and confirm its text distinguishes the declared bound from statistical detection."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","criteria":[{"criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"covered","evidence":["src/anomalyzer/early.py:22","src/anomalyzer/contracts.py:209","tests/test_capability_behaviors.py:74"],"notes":"Boundary values 10 and 20 are accepted; 9 and 21 violate the declared range."},{"criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"covered","evidence":["src/anomalyzer/early.py:25","tests/test_capability_behaviors.py:74"],"notes":"Parameterized tests distinguish equality to the threshold from strict exceedance for both change rules."},{"criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"covered","evidence":["src/anomalyzer/early.py:29","src/anomalyzer/early.py:53","tests/test_capability_behaviors.py:87"],"notes":"Zero/missing predecessors suppress relative comparisons, and missing values reset comparison history."},{"criterion":"Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them.","status":"covered","evidence":["src/anomalyzer/early.py:34","src/anomalyzer/early.py:121","tests/test_agentic.py:49","tests/test_cli.py:134"],"notes":"Violation records retain rule ID/kind/bounds; default eligibility is false and replay requires explicit permission."}],"ignore_findings":[],"done":true,"input_fingerprint":"60d24e84faee9c1e2ae8f9337d17334b37c89305093404da2e84c51526111d0b","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Apply my stated limits"},"impact":1,"gaps":0,"r":43,"x":1086.6666666666667,"y":406},{"id":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","label":"Find changes in conversion...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","acceptance":["Ratios retain the computed value, formula, and original source indexes for each sample.","Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","Derived series can be analyzed with the frozen or adaptive single-resolution recipe."],"guidance":["Relationships are caller-supplied declarations, not discovered or causal relationships.","A difference requires identical declared units; a normalized residual requires a positive supplied scale.","A joint condition is a declared 0/1 comparison, not statistical proof."],"path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","dependencies":["relationships/quality/require-compatible-sources","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_examples.py"],"automatedChecks":[{"id":"ratio-lineage","description":"Explain a changing rate with the exact source indexes and formula.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"other-comparisons","description":"Check declared differences, normalized residuals, and combined conditions.","command":"python -m pytest -q tests/test_agentic.py::test_remaining_relationship_primitives"},{"id":"unavailable-values","description":"Keep zero-denominator and missing-source observations unavailable.","command":"python -m pytest -q tests/test_agentic.py::test_missing_and_zero_denominator_are_unavailable_not_zero"},{"id":"adaptive-relationship","description":"Use the adaptive recipe for a declared relationship.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_recipe_runs_for_declared_relationships"}],"manualChecks":["Inspect examples/agentic/conversion.json and distinguish its relationship departure from the individual source metrics."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","criteria":[{"criterion":"Ratios retain the computed value, formula, and original source indexes for each sample.","status":"covered","evidence":["src/anomalyzer/relationships.py:84","src/anomalyzer/relationships.py:152","tests/test_agentic.py:211"],"notes":"Assertions check exact ratio values, formula, and both original source indexes."},{"criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"covered","evidence":["src/anomalyzer/relationships.py:97","src/anomalyzer/relationships.py:134","tests/test_agentic.py:285"],"notes":"Parameterized assertions independently specify difference, normalized residual, all-condition, and any-condition outputs."},{"criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"covered","evidence":["src/anomalyzer/relationships.py:89","src/anomalyzer/relationships.py:142","tests/test_agentic.py:226"],"notes":"Lineage retains null values with missing_source_value/zero_denominator reasons. Explicit zero_denominator=error intentionally fails instead, as documented."},{"criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:87","src/anomalyzer/orchestrator.py:135","tests/test_agentic.py:171","tests/test_examples.py:17"],"notes":"Both supported recipes dispatch derived datasets through the shared single-resolution engine."}],"ignore_findings":[],"done":true,"input_fingerprint":"7fa8f7ffdc0d852ad961b3e1753761d1bd3b00b87b8c001a957ec19e26d068b9","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Compare ratios and metric gaps"},"impact":1,"gaps":0,"r":43,"x":1270,"y":406},{"id":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","label":"Find unusual spikes and dr...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","acceptance":["A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","Separated point departures remain separate episodes, and positional episode intervals use integer coordinates."],"guidance":["Detection is a numerical departure from the selected expectation, not a claim about cause or business impact."],"path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/evidence.py","tests/test_analysis.py"],"automatedChecks":[{"id":"point-evidence","description":"Detect an injected spike and retain deterministic earlier numerical evidence.","command":"python -m pytest -q tests/test_analysis.py::test_spike_and_no_lookahead"},{"id":"adjacent-run","description":"Describe adjacent point triggers and their contributing positions.","command":"python -m pytest -q tests/test_analysis.py::test_anomaly_stream_detects_consecutive_run"},{"id":"isolated-events","description":"Keep separated anomalies from becoming a consecutive-run pattern.","command":"python -m pytest -q tests/test_analysis.py::test_isolated_anomalies_do_not_form_second_order_pattern"},{"id":"positional-episodes","description":"Use integer intervals for position-only episode evidence.","command":"python -m pytest -q tests/test_analysis.py::test_positional_episode_uses_integer_interval"}],"manualChecks":["Inspect observations and triggering_samples for examples/spike.json and distinguish episode grouping from pattern findings."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","criteria":[{"criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"covered","evidence":["src/anomalyzer/recipes.py:344","src/anomalyzer/recipes.py:388","tests/test_analysis.py:441"],"notes":"The injected spike retains its observed value, expectation, standardized residual, and point trigger."},{"criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"covered","evidence":["src/anomalyzer/baseline.py:31","src/anomalyzer/recipes.py:174","tests/test_analysis.py:441","tests/test_trend.py:36"],"notes":"Assertions establish repeatability and unchanged earlier evidence under later evaluation changes with fixed seasonal settings. Automatic positional season discovery has the separately documented streaming limitation."},{"criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"covered","evidence":["src/anomalyzer/evidence.py:95","tests/test_analysis.py:274"],"notes":"The test checks one three-sample run, exact contributing positions, direction, and first detection."},{"criterion":"Separated point departures remain separate episodes, and positional episode intervals use integer coordinates.","status":"covered","evidence":["src/anomalyzer/evidence.py:12","src/anomalyzer/evidence.py:23","tests/test_analysis.py:263","tests/test_analysis.py:369"],"notes":"Separated spikes produce two episodes without a consecutive-run pattern; positional evidence produces interval [9, 9]."}],"ignore_findings":[],"done":true,"input_fingerprint":"062833cc1bd19d5734b0f3c63d1e81666ffb568ead3399e359d62db2c0c0677d","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Find spikes and drops"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","label":"Know when the expected bas...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","acceptance":["Repeating seasonal structure with stable calibration can yield supported readiness.","Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","Insufficient calibration support yields not_assessed, including when the run otherwise completes.","Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment."],"guidance":["Readiness does not prove real-world effectiveness and is not an automatic action-policy gate."],"path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/readiness.py","src/anomalyzer/recipes.py","tests/test_readiness.py"],"automatedChecks":[{"id":"supported-reference","description":"Support repeated seasonality and preserve readiness under later changes.","command":"python -m pytest -q tests/test_readiness.py::test_repeating_season_and_trend_support_reference"},{"id":"weak-season","description":"Warn on unstable seasonality even when no findings occur.","command":"python -m pytest -q tests/test_readiness.py::test_unstable_declared_season_warns_without_calling_series_inapplicable tests/test_readiness.py::test_documented_random_example_has_no_findings_but_warns_on_reference"},{"id":"calibration-drift","description":"Warn when residual calibration drifts despite a repeated training season.","command":"python -m pytest -q tests/test_readiness.py::test_calibration_drift_warns_even_when_training_season_repeats"},{"id":"not-assessed","description":"Distinguish uncalibrated history and too few calibration samples.","command":"python -m pytest -q tests/test_readiness.py::test_readiness_is_not_assessed_without_calibrated_evaluation tests/test_readiness.py::test_short_calibration_is_not_assessed_even_when_analysis_completes"}],"manualChecks":["Read readiness reasons alongside findings; a lack of findings alone does not establish a trustworthy reference."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","criteria":[{"criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"covered","evidence":["src/anomalyzer/readiness.py:128","tests/test_readiness.py:8"],"notes":"Assertions establish repeated seasonality, strong calibration improvement, and supported readiness."},{"criterion":"Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","status":"covered","evidence":["src/anomalyzer/readiness.py:113","src/anomalyzer/readiness.py:122","tests/test_readiness.py:28","tests/test_readiness.py:41"],"notes":"Completed analyses retain caution reasons for weak seasonal repetition or calibration-center drift."},{"criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"covered","evidence":["src/anomalyzer/readiness.py:77","tests/test_readiness.py:56","tests/test_readiness.py:63"],"notes":"Both uncalibrated history and a completed run with three calibration samples return not_assessed."},{"criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"covered","evidence":["src/anomalyzer/recipes.py:327","tests/test_readiness.py:8"],"notes":"Readiness reads training/calibration inputs once; tests compare identical segment and aggregate assessments after a later shift."}],"ignore_findings":[],"done":true,"input_fingerprint":"d01367627e058ea39c78ee0e3616061a2e0725d06e8acd0b84b46a9655431e57","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Assess baseline quality"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","label":"Follow evolving growth whi...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","acceptance":["Each rolling expectation uses only observations before the sample being scored.","Persistent growth changes can be followed more closely than by the frozen default recipe.","An isolated extreme value is held out of future model references rather than creating a seasonal echo.","Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected."],"guidance":["A persistent third extreme miss in one direction may be admitted to the rolling reference; this is not a manual regime reset."],"path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"causal-window","description":"Keep a known rolling expectation invariant when future values are appended.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_expectation_is_one_step_and_prefix_invariant"},{"id":"changing-growth","description":"Follow a persistent growth change and report model admissions.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_tracks_a_persistent_growth_change"},{"id":"isolated-extreme","description":"Exclude an isolated extreme from later reference estimates.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_contains_an_isolated_extreme"},{"id":"bounded-settings","description":"Reject conflicting or unbounded adaptive settings.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_settings_reject_conflicting_or_unbounded_trend"}],"manualChecks":["Inspect the adaptive recipe's rolling-window parameters and frozen calibration before interpreting its final expectation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","criteria":[{"criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"covered","evidence":["src/anomalyzer/baseline.py:61","tests/test_analysis.py:48"],"notes":"All rolling-reference positions precede the scored index. The independent expectation remains unchanged after future values are appended."},{"criterion":"Persistent growth changes can be followed more closely than by the frozen default recipe.","status":"covered","evidence":["src/anomalyzer/recipes.py:145","src/anomalyzer/recipes.py:370","tests/test_analysis.py:58"],"notes":"The test requires adaptive held-out MAE below half the frozen recipe's MAE and a close final expectation."},{"criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","tests/test_analysis.py:81"],"notes":"The isolated spike is excluded from model references; its next seasonal expectation closely matches the control. Persistent third misses are admitted as guidance explicitly permits."},{"criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:41","tests/test_analysis.py:104"],"notes":"Both incompatible explicit trend selection and excessive slope lookback raise specific validation errors."}],"ignore_findings":[],"done":true,"input_fingerprint":"984422f1fd3ab3d22cf7b4ba9d0bd06629fc228a8e53d7ad44439cd8d35ec0db","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Follow evolving growth"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","label":"Account for daily and week...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"planned","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","acceptance":["Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","Insufficient support for a requested cycle produces an explicit limitation or declared fallback rather than a silently fitted component.","Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle."],"guidance":["Begin with a bounded daily-and-weekly use case; overlapping cycles share information and cannot be treated as independent evidence.","Proposed automated checks cover combined ordinary cycles, future isolation, unsupported cycle history, and extreme-value influence across cycles.","Manual review should inspect a weekday morning spike and verify the baseline explains ordinary calendar variation without explaining away the spike."],"path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Represent more than one repeating cycle"},"impact":0,"gaps":3,"r":47,"x":170,"y":406},{"id":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","label":"Limit how extreme values d...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","acceptance":["Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","Training and calibration diagnostics identify excluded extreme observations.","An extreme evaluated sample remains evidence while its expected value is used in future references.","Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior."],"guidance":["Protection keeps the observed evidence; it does not delete observations or certify that a departure was an incident."],"path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/robust.py","src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py","tests/test_cli.py"],"automatedChecks":[{"id":"robust-reference","description":"Limit an extreme weekly reference and use earlier references at lag one.","command":"python -m pytest -q tests/test_analysis.py::test_robust_weekly_reference_limits_one_high_week tests/test_analysis.py::test_robust_lag_one_uses_prior_seasonal_values"},{"id":"training-calibration","description":"Record training and calibration exclusions and retain a useful growth fit.","command":"python -m pytest -q tests/test_trend.py::test_robust_training_fit_and_calibration_exclusions"},{"id":"include-choice","description":"Compare robust and include modes through the CLI and show their differing triggers.","command":"python -m pytest -q tests/test_cli.py::test_outlier_handling_cli"}],"manualChecks":["Inspect excluded_from_model, reference_action, and outlier-handling diagnostics alongside the original observed value."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","criteria":[{"criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"covered","evidence":["src/anomalyzer/baseline.py:39","tests/test_analysis.py:25"],"notes":"Assertions check a independently computed weighted reference and invariance when an already-trimmed high value becomes more extreme."},{"criterion":"Training and calibration diagnostics identify excluded extreme observations.","status":"covered","evidence":["src/anomalyzer/trend.py:178","src/anomalyzer/recipes.py:287","tests/test_trend.py:65"],"notes":"Tests identify exact training/calibration exclusions and compare their influence on trend selection and residual scale."},{"criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","src/anomalyzer/recipes.py:392","tests/test_analysis.py:263","tests/test_trend.py:11"],"notes":"The observed departure and trigger survive; only reference_values receives the prior expectation, preventing the next seasonal echo."},{"criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"covered","evidence":["src/anomalyzer/recipes.py:151","src/anomalyzer/recipes.py:475","tests/test_cli.py:95"],"notes":"CLI assertions distinguish robust triggers {65} from include-mode triggers {65, 72}."}],"ignore_findings":[],"done":true,"input_fingerprint":"183b8696fdc9786878b3e54c4a9e5f68180b47e157528bef953274a3487d49a7","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Protect later comparisons"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","label":"Choose the repeating cycle...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","acceptance":["An explicit season length is honored and reported as explicit.","A repeating positional series can infer a seasonal lag and report the selection window.","A positional series without a sufficiently stable candidate falls back to lag one and records why.","The reserved calibration and evaluation tail does not determine positional season selection."],"guidance":["Candidate discovery uses the input prefix before the reserved tail and confirms the candidate in an early window. It is not fully prefix-invariant streaming discovery.","Supply season_length explicitly when stable streaming season selection is required."],"path":".capabilities/analysis/expectations/select-season-length.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/seasonality.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"declared-season","description":"Honor a separate explicit season setting.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"inference-fallback","description":"Infer a repeating lag or record lag-one fallback when no season is supported.","command":"python -m pytest -q tests/test_analysis.py::test_positional_season_inference_and_fallback"},{"id":"reserved-tail","description":"Keep selection unchanged when only the reserved tail changes.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_season_inference_excludes_reserved_tail"},{"id":"cli-inference","description":"Expose the inferred lag and its source for bare JSON values in the CLI.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"}],"manualChecks":["Review selection_window and ranked candidates for a positional series before treating its inferred season as an ongoing monitoring assumption."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","criteria":[{"criterion":"An explicit season length is honored and reported as explicit.","status":"covered","evidence":["src/anomalyzer/recipes.py:55","tests/test_analysis.py:234"],"notes":"Explicit field presence bypasses inference, and the test checks the reported source."},{"criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"covered","evidence":["src/anomalyzer/seasonality.py:96","tests/test_analysis.py:247"],"notes":"The repeating seven-position series selects lag 7 and reports confirmation window [0, 27]."},{"criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"covered","evidence":["src/anomalyzer/seasonality.py:79","src/anomalyzer/seasonality.py:84","tests/test_analysis.py:247"],"notes":"Candidate-strength and early-confirmation failures return lag 1 with a reason; the quadratic example verifies fallback."},{"criterion":"The reserved calibration and evaluation tail does not determine positional season selection.","status":"covered","evidence":["src/anomalyzer/seasonality.py:69","tests/test_capability_behaviors.py:147"],"notes":"Discovery excludes calibration_size plus one reserved sample. Changing that tail preserves the complete inference diagnostics; arbitrary appended-prefix invariance is explicitly outside the guidance."}],"ignore_findings":[],"done":true,"input_fingerprint":"3317b1727dc3cabfcfc9b87808d9249d0d59f111b8fd5f8a1fad653b21343980","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Choose a repeating cycle"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","label":"Interpret rate changes wit...","area":"analysis","scopes":["analysis","analysis/rates"],"scope":"analysis/rates","scopeLabel":"analysis/rates","status":"planned","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","acceptance":["Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","Zero denominators remain unavailable, and low support is explicitly marked; intervals remain within the valid zero-to-one rate domain.","Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations."],"guidance":["Independence and stable exposure are modeling assumptions; clustered events, overdispersion, or changing populations require explicit limitations.","Proposed automated checks cover zero denominators, tiny counts, zero-or-all-success boundaries, and equal rates at different volumes.","Manual review should compare sparse and high-volume conversion changes and confirm the explanation discloses assumptions without inventing incident probability."],"path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Judge rates using their exposure"},"impact":0,"gaps":3,"r":47,"x":194,"y":524},{"id":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","label":"See changes before the day...","area":"analysis","scopes":["analysis","analysis/resolution"],"scope":"analysis/resolution","scopeLabel":"analysis/resolution","status":"implemented","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","acceptance":["Callers can select sum, mean, or last aggregation for completed and partial periods.","Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","An explicitly unfinished source sample is excluded from both analysis views.","A source cadence that does not evenly divide 24 hours is rejected."],"guidance":["Periods are fixed 24-hour windows, not daylight-saving-aware local calendar days.","Native and daily views share source observations and are correlated.","Legacy schema 1.0 requires complete input and does not support resets. Schema 1.1 labels sparse native/completed-day views and preserves reset boundaries; daily sums require declared additive volume and fully observed children."],"path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","dependencies":["data/quality/exclude-incomplete-periods","analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"selected-aggregation","description":"Check sum, mean, and last for completed days and trailing partial aggregates.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_multi_resolution_uses_selected_aggregation"},{"id":"two-views","description":"Return completed days, native evidence, and a linked trailing partial-day marker.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_current_period_and_keeps_two_views"},{"id":"unfinished-source","description":"Exclude a marked incomplete source sample from both views.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days"},{"id":"dividing-cadence","description":"Reject subperiod durations that do not evenly divide a day.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_rejects_non_dividing_frequency"}],"manualChecks":["Inspect the partial aggregate's observed_aggregate and intraday_evidence_refs together with the two method namespaces."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","criteria":[{"criterion":"Callers can select sum, mean, or last aggregation for completed and partial periods.","status":"covered","evidence":["src/anomalyzer/contracts.py:18","src/anomalyzer/recipes.py:539","tests/test_capability_behaviors.py:178"],"notes":"Parameterized assertions check independently known completed and partial values for all three aggregation choices."},{"criterion":"Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","status":"covered","evidence":["src/anomalyzer/recipes.py:617","src/anomalyzer/recipes.py:667","src/anomalyzer/recipes.py:680","tests/test_analysis.py:170"],"notes":"Separate datasets, configurations, and namespaces preserve the two correlated views."},{"criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"covered","evidence":["src/anomalyzer/recipes.py:639","src/anomalyzer/recipes.py:690","tests/test_analysis.py:170"],"notes":"The marker contains finalized progress, observed aggregate, native evidence references, and an explicit exclusion flag."},{"criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"covered","evidence":["src/anomalyzer/validation.py:44","src/anomalyzer/recipes.py:564","tests/test_analysis.py:200"],"notes":"Both child datasets use prepared finalized observations; tests verify counts and retention of the excluded original value."},{"criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"covered","evidence":["src/anomalyzer/recipes.py:572","tests/test_analysis.py:221"],"notes":"A five-hour cadence raises the specific dividing-cadence error."}],"ignore_findings":[],"done":true,"input_fingerprint":"cdcdea1166a4e3d2523b594bec938142135e17a6b4279732e52819a486d1b4f2","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Compare finalized subperiods"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","label":"Compare days and weeks in ...","area":"data","scopes":["data","data/calendar"],"scope":"data/calendar","scopeLabel":"data/calendar","status":"planned","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","acceptance":["Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","Periods report UTC boundaries and local labels; daylight-saving days can contain 23 or 25 hours without being forced into fixed 24-hour windows.","Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation."],"guidance":["Calendar grouping needs a distinct contract from fixed elapsed cadence; it must not silently reinterpret existing frequency or aggregation settings.","Proposed automated checks cover spring transitions, fall transitions, week boundaries, and missing native slots in a calendar period.","Manual review should inspect a daylight-saving week and confirm local labels, observed coverage, and completed-versus-open comparisons are understandable."],"path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Declare local calendar boundaries"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","label":"Analyze many entities with...","area":"data","scopes":["data","data/cohorts"],"scope":"data/cohorts","scopeLabel":"data/cohorts","status":"planned","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","acceptance":["Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","Training, calibration, relationships, and cases do not cross entity boundaries unless a separate explicit aggregate comparison is requested.","The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident."],"guidance":["Batch orchestration should preserve existing per-request contracts rather than silently lifting schema limits; multiple-testing limits stay visible.","Proposed automated checks cover partition isolation, batch-versus-single equivalence, one-entity failure, and shared resource-budget exhaustion.","Manual review should inspect a mixed batch of healthy, changed, and sparse entities and confirm the summary does not equate missing support with health."],"path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Keep each entity's evidence separate"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","label":"Know when missing updates ...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"planned","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","acceptance":["Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","Results distinguish current, late, missing, and not-assessed freshness and expose the last observed update and the deadline used.","A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision."],"guidance":["Freshness depends on declared collection expectations; it does not prove source completeness and does not start a scheduler or send a notification.","Proposed automated checks cover exact deadlines, allowed lateness, absent source updates, and stale-source propagation to relationships.","Manual review should compare a stale feed with a true metric drop and confirm the report distinguishes collection uncertainty from numerical departure."],"path":".capabilities/data/quality/report-source-freshness.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Check whether the sources are current"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","label":"See how late or corrected ...","area":"data","scopes":["data","data/revisions"],"scope":"data/revisions","scopeLabel":"data/revisions","status":"planned","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","acceptance":["A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","Results preserve prior and new evidence versions and identify findings that were added, changed, or withdrawn with traceable source lineage.","Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results."],"guidance":["Schema-1.1 snapshot revision selection and as-of availability are implemented in data/quality/analyze-sparse-observations; durable prior/new evidence diffing and automatic reconciliation in this capability remain planned.","A corrected event-time result does not replace the record of what was known earlier; callers control retention and any notification about a revision.","Proposed automated checks cover late completion, corrected calibration data, downstream relationship changes, and conflicting source versions.","Manual review should inspect a drop withdrawn after ingestion completes and confirm both the original limitation and the revision explanation are retained."],"path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Reconcile revisions without losing history"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","label":"Choose settings using held...","area":"evaluation","scopes":["evaluation","evaluation/comparisons"],"scope":"evaluation/comparisons","scopeLabel":"evaluation/comparisons","status":"planned","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","acceptance":["Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","The output preserves every candidate's settings and evaluation provenance; adopting a configuration requires an explicit caller decision."],"guidance":["Bounded comparison is not unrestricted parameter search; held-out results describe the supplied history and do not guarantee future effectiveness.","Proposed automated checks cover split isolation, comparable replay inputs, failed candidate retention, and unchanged active settings after comparison.","Manual review should compare an earlier but noisier configuration with a quieter one and confirm the tradeoff is visible without an unsupported winner."],"path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Review the tradeoff before adopting settings"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","label":"Measure detection timing a...","area":"evaluation","scopes":["evaluation","evaluation/outcomes"],"scope":"evaluation/outcomes","scopeLabel":"evaluation/outcomes","status":"planned","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","acceptance":["Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","Timing reports distinguish the incident's supplied start from the first available candidate, calibrated detection, and policy-eligible evidence.","Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown."],"guidance":["Detection timing does not measure response effectiveness or establish that an unmatched finding was false; incident labels are caller judgments.","Proposed automated checks cover boundary matching, ambiguous overlaps, incomplete label coverage, and separate candidate versus eligibility timing.","Manual review should inspect a matched and an unmatched episode and confirm the reported timing and accuracy limits agree with the supplied label coverage."],"path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Compare detection with reviewed outcomes"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","label":"Replay history as it would...","area":"evaluation","scopes":["evaluation","evaluation/replay"],"scope":"evaluation/replay","scopeLabel":"evaluation/replay","status":"planned","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","acceptance":["Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","Repeating a replay preserves numerical evidence and first-detection timing apart from generated IDs and runtime; stopped replays retain completed steps."],"guidance":["Event-time replay does not reconstruct arrival-time availability unless the caller supplies versioned arrival history.","Proposed automated checks cover future-tail isolation, short-history cutoffs, deterministic repeated replay, and interrupted replay retention.","Manual review should inspect one change from its first available observation through calibration and confirm the explanation uses only then-known evidence."],"path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Reconstruct the evidence available at the time"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","label":"Update evidence when the n...","area":"execution","scopes":["execution","execution/incremental"],"scope":"execution/incremental","scopeLabel":"execution/incremental","status":"planned","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","acceptance":["Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","Supported explicitly configured recipes produce the same numerical evidence as full batch analysis of the combined history, including cross-chunk patterns.","Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations."],"guidance":["Start with explicit season lengths; input-prefix season discovery is not currently guaranteed to be invariant when the history grows.","Proposed automated checks cover chunk-versus-batch equivalence, cross-chunk CUSUM, idempotent retry, and failed-append state preservation.","Manual review should follow a sustained change across several append batches and confirm detection timing and provisional evidence remain understandable."],"path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Analyze the next finalized samples"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","label":"Keep useful evidence when ...","area":"execution","scopes":["execution","execution/resilience"],"scope":"execution/resilience","scopeLabel":"execution/resilience","status":"implemented","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","acceptance":["A method failure preserves evidence completed before the failure and reports the error.","Runtime budgets are checked cooperatively and produce an explicit stop reason.","CLI byte limits and execution failures produce distinct error exits.","A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial."],"guidance":["Default limits are 10,000 source observations per request, 10 MB CLI input, and 60 seconds cooperative runtime.","Numerical results are deterministic for fixed declared settings; generated run IDs and runtimes vary."],"path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/budget.py","src/anomalyzer/recipes.py","src/anomalyzer/orchestrator.py","src/anomalyzer/io.py","src/anomalyzer/cli.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"method-and-budget","description":"Retain pre-failure evidence and report a runtime-budget stop.","command":"python -m pytest -q tests/test_analysis.py::test_failure_and_budget"},{"id":"cli-limits","description":"Enforce byte limits and distinguish execution failure in CLI output and exit status.","command":"python -m pytest -q tests/test_cli.py::test_size_limit_and_execution_exit"},{"id":"partial-siblings","description":"Preserve healthy analysis when a sibling dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Inspect run status, stop_reason, and each target's status before relying on a result; confirm limits are presented as cooperative rather than hard process timeouts."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","criteria":[{"criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"covered","evidence":["src/anomalyzer/recipes.py:400","src/anomalyzer/recipes.py:448","tests/test_analysis.py:509"],"notes":"The failure regression preserves evidence through index 44 and reports the injected error at index 45."},{"criterion":"Runtime budgets are checked cooperatively and produce an explicit stop reason.","status":"covered","evidence":["src/anomalyzer/budget.py:18","src/anomalyzer/recipes.py:108","src/anomalyzer/recipes.py:508","tests/test_analysis.py:509"],"notes":"Cooperative target/sample checks expose runtime_budget and budget_exceeded; this is not a hard process timeout."},{"criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:429","tests/test_cli.py:108"],"notes":"The full suite executed the byte-limit and execution-exit regression successfully; invalid input exits 2 and execution failure exits 1."},{"criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:125","src/anomalyzer/orchestrator.py:171","src/anomalyzer/orchestrator.py:185","tests/test_capability_behaviors.py:159","tests/test_capability_behaviors.py:30"],"notes":"Healthy dataset evidence survives target-specific failure; inspected incomplete-usage reports partial with no invented healthy relationship."}],"ignore_findings":[],"done":true,"input_fingerprint":"146a9c8396fd7a2d039f75077dc0b82a0778685d20c3782aa284d5f0b6ed94be","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect incomplete execution"},"impact":0,"gaps":0,"r":38,"x":720,"y":642},{"id":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","label":"Resume local analysis afte...","area":"execution","scopes":["execution","execution/state"],"scope":"execution/state","scopeLabel":"execution/state","status":"planned","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","acceptance":["A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","Restoring a supported state and continuing yields the same numerical evidence and detection positions as uninterrupted processing of the same observations.","Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled."],"guidance":["Stored observations and context can contain sensitive data; document exactly what a snapshot contains and avoid automatic persistence.","Proposed automated checks cover restart equivalence, corrupt snapshots, version or settings mismatch, and output overwrite protection.","Manual review should inspect a snapshot and resumed explanation and confirm their provenance and retained-data scope are clear to an application owner."],"path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Preserve state across restarts"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","label":"Understand findings in a r...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","acceptance":["Point-anomaly episode counts and residual-pattern findings are labeled separately.","Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","Readiness status remains visible, including when it was not assessed."],"guidance":["Readable summaries should retain the evidence's claim limits and allow drilldown into JSON."],"path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/cli.py","tests/test_cli.py","examples/agentic/conversion.json","examples/agentic/unit-cost.json"],"automatedChecks":[{"id":"episodes-patterns","description":"Distinguish shift patterns from point episodes in readable output.","command":"python -m pytest -q tests/test_cli.py::test_readable_output_distinguishes_episodes_from_patterns"},{"id":"relationship-case","description":"Explain a conversion case with entity, units, comparison, and a suggested check.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_relationship_request_through_cli"},{"id":"explicit-rule","description":"Describe the unit-cost rule boundary and unassessed readiness.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the conversion and unit-cost example output and confirm the descriptions state the supported numerical comparison in plain language."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","criteria":[{"criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"covered","evidence":["src/anomalyzer/cli.py:347","src/anomalyzer/cli.py:405","tests/test_cli.py:143"],"notes":"Readable tests distinguish zero-episode residual shifts from a spike episode and retain the overlap warning."},{"criterion":"Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","status":"covered","evidence":["src/anomalyzer/cli.py:291","src/anomalyzer/cli.py:308","src/anomalyzer/cli.py:331","tests/test_cli.py:117","examples/agentic/conversion.json"],"notes":"Conversion output identifies site=demo, observed/expected orders_per_visit, the relationship, and a collection-completeness investigation question."},{"criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"covered","evidence":["src/anomalyzer/cli.py:318","tests/test_cli.py:134","examples/agentic/unit-cost.json"],"notes":"Unit-cost output names review-above-13c and its [0, 0.13] boundary as a declared rule crossing."},{"criterion":"Readiness status remains visible, including when it was not assessed.","status":"covered","evidence":["src/anomalyzer/cli.py:352","src/anomalyzer/cli.py:399","tests/test_cli.py:134"],"notes":"Readable output includes not_assessed and its supporting reasons without claiming measured alert accuracy."}],"ignore_findings":[],"done":true,"input_fingerprint":"98108868b000972fd9b3c3f082846fbd3f3b08c399ef7ed424c23245d6935fef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Read the findings"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","label":"Share an investigation tha...","area":"interfaces","scopes":["interfaces","interfaces/export"],"scope":"interfaces/export","scopeLabel":"interfaces/export","status":"planned","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","acceptance":["Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","Source observations and context are included only by explicit selection; omitted data and resulting limits on independent reproduction are disclosed."],"guidance":["Fingerprints identify included provenance; they do not prove source truth. A portable report must make privacy scope and reproduction limits inspectable.","Proposed automated checks cover manifest consistency, partial-run export, omitted-source disclosure, and overwrite protection.","Manual review should open a bundle outside the original session and confirm a reviewer can trace one finding and see exactly what cannot be reproduced."],"path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Preserve a portable evidence record"},"impact":0,"gaps":3,"r":47,"x":927.3333333333334,"y":760},{"id":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","label":"Install a versioned analys...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","acceptance":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"guidance":["Guard owns scheduling, persistence, business severity, incidents and notifications.","A published package does not establish integration into UsageTap Guard."],"path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","interfaces/cli/analyze-files-and-streams","interfaces/mcp/connect-read-only-local-server"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["pyproject.toml","src/anomalyzer/_version.py","scripts/release.py","scripts/publish_release.py","scripts/verify_distribution.py","scripts/verify_lambda.sh","examples/library/guard_smoke.py",".github/workflows/ci.yml",".github/workflows/release.yml","docs/releases/README.md","tests/test_release.py"],"automatedChecks":[{"id":"release-safety","description":"Reject mismatched tags, altered assets and mutable publication.","command":"python -m pytest -q tests/test_release.py"},{"id":"installed-artifacts","description":"Install wheel and sdist outside the checkout and invoke public APIs.","command":"python scripts/verify_distribution.py --python 3.12"},{"id":"release-inventory","description":"Check archive contents, version metadata and SHA-256 inventory.","command":"python scripts/release.py verify"},{"id":"linux-offline-consumer","description":"Bundle and run with networking disabled in Linux Python 3.12.","command":"bash scripts/verify_lambda.sh"}],"manualChecks":["Follow the release guide and confirm Guard retains domain severity and incident ownership; inspect readiness and evidence limits before adopting."],"verificationGaps":[{"code":"declared-gap","message":"interfaces/python/install-versioned-library: First immutable GitHub Release requires owner setup and a reviewed version tag.","capabilityId":"interfaces/python/install-versioned-library","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\python\\install-versioned-library.capability.yaml"}],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for interfaces/python/install-versioned-library.","criteria":[{"criterion":"Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","status":"partial","evidence":["src/anomalyzer/_version.py:1","scripts/release.py:1","scripts/release.py:20","scripts/release.py:39","scripts/release.py:48","scripts/release.py:54","scripts/publish_release.py:19","scripts/publish_release.py:27","scripts/publish_release.py:29","scripts/publish_release.py:39","scripts/verify_lambda.sh:14","scripts/verify_lambda.sh:17","scripts/verify_lambda.sh:19","scripts/verify_lambda.sh:25","scripts/verify_lambda.sh:27",".github/workflows/ci.yml:14",".github/workflows/ci.yml:19",".github/workflows/ci.yml:27",".github/workflows/ci.yml:41",".github/workflows/ci.yml:46",".github/workflows/release.yml:28","docs/releases/README.md:12","docs/releases/README.md:13","docs/releases/README.md:14","docs/releases/README.md:17","docs/releases/README.md:18","tests/test_release.py:60","tests/test_release.py:71","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","status":"partial","evidence":["scripts/verify_distribution.py:1","scripts/verify_distribution.py:42","examples/library/guard_smoke.py:32","examples/library/guard_smoke.py:49","examples/library/guard_smoke.py:65","examples/library/guard_smoke.py:69","examples/library/guard_smoke.py:75",".github/workflows/ci.yml:57","docs/releases/README.md:13","docs/releases/README.md:208","docs/releases/README.md:211","docs/releases/README.md:213"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:34","scripts/release.py:81","scripts/release.py:82","scripts/release.py:85","scripts/release.py:92","scripts/publish_release.py:1","scripts/publish_release.py:25","scripts/publish_release.py:26","scripts/publish_release.py:27","scripts/publish_release.py:28","scripts/verify_distribution.py:41","scripts/verify_lambda.sh:24","examples/library/guard_smoke.py:39",".github/workflows/ci.yml:42",".github/workflows/ci.yml:45",".github/workflows/ci.yml:46",".github/workflows/ci.yml:49",".github/workflows/ci.yml:51",".github/workflows/release.yml:31",".github/workflows/release.yml:33",".github/workflows/release.yml:36",".github/workflows/release.yml:37","docs/releases/README.md:3","docs/releases/README.md:12","docs/releases/README.md:14","docs/releases/README.md:15","docs/releases/README.md:19","tests/test_release.py:1","tests/test_release.py:21","tests/test_release.py:23","tests/test_release.py:52","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:81","scripts/release.py:82","scripts/publish_release.py:40","scripts/verify_distribution.py:41","scripts/verify_lambda.sh:6","scripts/verify_lambda.sh:17","scripts/verify_lambda.sh:19",".github/workflows/ci.yml:30",".github/workflows/ci.yml:31",".github/workflows/ci.yml:51",".github/workflows/ci.yml:52",".github/workflows/ci.yml:90","docs/releases/README.md:12","docs/releases/README.md:15","docs/releases/README.md:58","docs/releases/README.md:59","docs/releases/README.md:75"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","partial: Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","partial: Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","partial: A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","name":"Python integration","label":"Python integration / Embed analysis in my application / Install a released dependency"},"impact":0,"gaps":1,"r":41,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","label":"Ask an agent to explain un...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","acceptance":["The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit."],"guidance":["This contract covers the supplied instructions and demonstrated analysis workflow; it does not guarantee that every agent host follows the instructions.","Synthetic examples demonstrate mechanics, not real-world detection accuracy."],"path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/agents/openai.yaml","src/anomalyzer/cli.py","src/anomalyzer/mcp.py","tests/test_cli.py","tests/test_agentic.py","tests/test_mcp.py"],"automatedChecks":[{"id":"structured-cli","description":"Exercise the documented structured CLI workflow and preserve incomplete-history evidence.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence tests/test_cli.py::test_bare_json_values_cli"},{"id":"declared-relationships","description":"Check explicit ratio semantics and unavailable source evidence.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"mcp-drilldown","description":"Follow compact analysis to a finding and its source evidence.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case"}],"manualChecks":["Follow the conversion example through the CLI and MCP drilldown; compare the Skill instructions with actual evidence and check that its proposed text separates calibrated findings, early candidates, readiness, and caller-owned actions."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","criteria":[{"criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:13",".agents/skills/anomaly-detection/SKILL.md:41","tests/test_cli.py:13"],"notes":"Executable resolution and JSON commands are explicit. Manual conversion and spike CLI runs use the installed numerical engine."},{"criterion":"The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:24",".agents/skills/anomaly-detection/SKILL.md:32","tests/test_cli.py:51","tests/test_agentic.py:211"],"notes":"Instructions preserve scope, units, missingness, declared relationships, and exact alignment; existing runtime checks establish these prerequisites."},{"criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:101",".agents/skills/anomaly-detection/SKILL.md:134",".agents/skills/anomaly-detection/agents/openai.yaml:4"],"notes":"The supplied instructions lead with plain-language changes and then qualify maturity/readiness/overlap. This establishes instructional content, not measured agent compliance."},{"criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:54",".agents/skills/anomaly-detection/SKILL.md:174","src/anomalyzer/mcp.py:169","tests/test_mcp.py:54"],"notes":"The documented three-tool workflow matches executed conversion drilldown, with session-only run IDs and explicit read-only/caller-owned delivery limits."}],"ignore_findings":[],"done":true,"input_fingerprint":"751723214aa340b1227ec347c22f130749fcdbfd49327c21b4c27ef68d7f0341","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Get an evidence-grounded explanation"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","label":"See the evidence in a char...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","acceptance":["The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes."],"guidance":["Native chart rendering and interactive figure features depend on the agent host; the checked-in fallback produces a static SVG for a single dataset.","A chart explains structured analysis; it does not replace the detector result."],"path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/scripts/quick_chart.py",".agents/skills/anomaly-detection/references/visualization.md","tests/test_skill_chart.py"],"automatedChecks":[{"id":"observed-evidence","description":"Render real observed and expected data and only actual point triggers.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_preserves_real_point_evidence"},{"id":"maturity-and-resets","description":"Keep provisional evidence non-triggering and label reviewed regimes.","command":"python -m pytest -q tests/test_skill_chart.py::test_provisional_scores_are_not_charted_as_triggers tests/test_skill_chart.py::test_quick_chart_marks_reviewed_resets"},{"id":"unsupported-scope","description":"Reject a multi-dataset result rather than mixing units.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_rejects_multiple_datasets"},{"id":"output-protection","description":"Require an explicit overwrite for existing charts.","command":"python -m pytest -q tests/test_skill_chart.py::test_chart_cli_requires_explicit_overwrite"}],"manualChecks":["Inspect a spike fallback chart and the detailed figure instructions; verify that axes, initialization context, trigger markers, and interpretation limits are clear, and that the Skill requests a chart only when the user asks."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","criteria":[{"criterion":"The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:189",".agents/skills/anomaly-detection/SKILL.md:226"],"notes":"The Skill explicitly requires a visual request after analysis and separates quick versus detailed rendering instructions."},{"criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:81",".agents/skills/anomaly-detection/scripts/quick_chart.py:59","tests/test_skill_chart.py:28","tests/test_skill_chart.py:43","tests/test_skill_chart.py:54"],"notes":"Real-result tests establish observed/expected lines, trigger-only markers, provisional non-triggering, units, initialization shading, and reset boundaries; the spike chart was inspected visually."},{"criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:85",".agents/skills/anomaly-detection/scripts/quick_chart.py:251","tests/test_skill_chart.py:63","tests/test_skill_chart.py:68"],"notes":"Multiple resolved datasets raise a specific error. The CLI test preserves an existing chart and replaces it only with explicit overwrite."},{"criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/references/visualization.md:31",".agents/skills/anomaly-detection/references/visualization.md:37",".agents/skills/anomaly-detection/references/visualization.md:45",".agents/skills/anomaly-detection/references/visualization.md:86"],"notes":"Detailed instructions require index alignment, maturity distinctions, first detection timing, visible unavailable status, and qualified causal claims; native interactive rendering remains host-dependent."}],"ignore_findings":[],"done":true,"input_fingerprint":"2780b6c3e363f89a30a0709772c0a46a9a96f1567fabb27e63934a3ca80dc982","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Inspect a requested chart"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","label":"Find which segments contri...","area":"investigation","scopes":["investigation","investigation/cohorts"],"scope":"investigation/cohorts","scopeLabel":"investigation/cohorts","status":"planned","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","acceptance":["Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","Segment ranking retains absolute signed contributions and baseline support; an arithmetic contributor is not automatically labeled anomalous or causal."],"guidance":["Ratios and overlapping customer groups are not additive totals; separate segment expectations must not be assumed to sum to a fitted total expectation.","Proposed automated checks cover exact reconciliation, offsetting segment changes, missing segments, and zero total departure.","Manual review should inspect a regional decline with an offsetting increase and confirm the explanation preserves signed contributions and evidence limits."],"path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Locate contributors to a total change"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","label":"Keep reviewed outcomes bes...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","acceptance":["A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","Review export distinguishes analyst judgment from detector output; annotation alone does not change calibration, eligibility, or regime boundaries."],"guidance":["Caller-supplied reviewer identity is attribution, not built-in authentication; storage, access control, response execution, and incident ownership stay with callers.","Proposed automated checks cover evidence-version linking, unresolved reviews, revision retention, and unchanged numerical evidence after annotation.","Manual review should inspect a disputed case with two review revisions and confirm judgments, uncertainties, and original findings remain distinguishable."],"path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Record the reviewed outcome"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","label":"Find earlier reviewed case...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","acceptance":["Search is bounded to caller-selected records and explicit compatibility filters; it does not scan unrelated local files or require a model API.","Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match."],"guidance":["Similar historical evidence does not establish the same cause or authorize repeating a past response; caller-selected records may be incomplete or biased.","Proposed automated checks cover compatibility filtering, bounded corpus access, missing comparison context, and no-match behavior.","Manual review should compare superficially similar cases with different outcomes and confirm precedent is presented as investigation context."],"path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Inspect comparable reviewed history"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","label":"Change action eligibility ...","area":"investigation","scopes":["investigation","investigation/policy"],"scope":"investigation/policy","scopeLabel":"investigation/policy","status":"implemented","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","acceptance":["Default policy leaves action and notification eligibility disabled.","Explicit-rule and calibrated-departure eligibility are controlled independently.","Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged."],"guidance":["Eligibility does not send notifications, execute actions, check readiness, or confer authority to act.","Monitoring schedules, delivery, reviewed incident state, and deduplication belong to the caller."],"path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","dependencies":["investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/agent.py","src/anomalyzer/cases.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"independent-notifications","description":"Replay notification and action choices without changing numerical evidence or the original result.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_separates_action_from_notification"},{"id":"calibrated-choice","description":"Enable calibrated departures only with the corresponding policy choice.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_enables_calibrated_departures_only_when_requested"},{"id":"nontriggering-candidates","description":"Keep descriptive candidates ineligible even under a permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"}],"manualChecks":["Inspect the unit-cost example under default and allow_explicit_rules policies and compare eligibility with the unchanged numerical violation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","criteria":[{"criterion":"Default policy leaves action and notification eligibility disabled.","status":"covered","evidence":["src/anomalyzer/contracts.py:329","tests/test_capability_behaviors.py:74"],"notes":"Default flags disable both eligibility types, including after replay resets policy."},{"criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"covered","evidence":["src/anomalyzer/agent.py:47","tests/test_capability_behaviors.py:123"],"notes":"Explicit-only permission does not enable a calibrated departure; the matching calibrated choice does."},{"criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"covered","evidence":["src/anomalyzer/agent.py:53","tests/test_capability_behaviors.py:101","tests/test_capability_behaviors.py:134"],"notes":"Notification-only replay enables a met rule while keeping action disabled; permissive settings still cannot enable candidates."},{"criterion":"Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged.","status":"covered","evidence":["src/anomalyzer/agent.py:42","src/anomalyzer/cases.py:57","src/anomalyzer/cases.py:105","tests/test_capability_behaviors.py:101","tests/test_agentic.py:49"],"notes":"Assertions compare unchanged original serialization, methods, evidence references, stable IDs, and eligibility-dependent revision changes."}],"ignore_findings":[],"done":true,"input_fingerprint":"ee67bcc1ec1380b8d84a0556d838d648ae6cef3f4f9b9e34e1b71e90eb00aeae","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Review action eligibility"},"impact":0,"gaps":0,"r":38,"x":1086.6666666666667,"y":878},{"id":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","label":"Review a possible new base...","area":"investigation","scopes":["investigation","investigation/regimes"],"scope":"investigation/regimes","scopeLabel":"investigation/regimes","status":"planned","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","acceptance":["A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","Producing a proposal leaves the active regime and original findings unchanged.","A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect."],"guidance":["A proposed boundary is not a confirmed cause or a retrospectively available detection; adaptive reference updates remain a separate behavior.","Proposed automated checks cover unchanged active results, short new regimes, rejected proposals, and accepted explicit-reset equivalence.","Manual review should inspect a persistent outage-like drop and a known launch and confirm neither becomes ordinary behavior merely because it persists."],"path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Review a proposed regime transition"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","label":"Distinguish changing perfo...","area":"relationships","scopes":["relationships","relationships/cohorts"],"scope":"relationships/cohorts","scopeLabel":"relationships/cohorts","status":"planned","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","acceptance":["Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","The report preserves aggregate and cohort rates, population weights, and the declared decomposition formula and reference-period convention.","Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes."],"guidance":["Decomposition is descriptive accounting; it does not estimate an intervention effect or establish that the comparison populations are causally equivalent.","Proposed automated checks cover stable rates with changing mix, stable mix with changing rates, offsetting cohort rates, and new or empty cohorts.","Manual review should inspect a Simpson's-paradox example and confirm the aggregate movement and cohort evidence can both be understood correctly."],"path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Separate performance changes from population mix"},"impact":0,"gaps":3,"r":47,"x":1270,"y":878},{"id":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","label":"Check whether a metric fol...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","acceptance":["A supplied lag and coefficient use the predictor at the declared earlier sample position.","Derived values retain response and predictor source indexes and source cutoffs in lineage.","When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","Samples lacking the earlier predictor or awaiting coefficient training remain unavailable."],"guidance":["Lag is measured in samples and is never searched automatically.","Exactly one of coefficient or training_size is required."],"path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","dependencies":["relationships/quality/require-compatible-sources"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_examples.py","examples/agentic/lagged-response.json"],"automatedChecks":[{"id":"supplied-response","description":"Use the declared lag and coefficient with correct predictor indexes.","command":"python -m pytest -q tests/test_agentic.py::test_lagged_response_uses_supplied_coefficient_and_prior_sample"},{"id":"frozen-coefficient","description":"Keep fitted coefficient and earlier lineage invariant when future samples arrive.","command":"python -m pytest -q tests/test_agentic.py::test_estimated_lag_coefficient_is_frozen_and_prefix_invariant"},{"id":"delayed-example","description":"Run the checked-in lagged-response demonstration against expected results.","command":"python -m pytest -q tests/test_examples.py::test_agentic_demo_checks_all_cases"},{"id":"pending-training","description":"Retain unavailable lineage until the full declared training prefix arrives, then freeze its coefficient.","command":"python -m pytest -q tests/test_agentic.py::test_lag_fit_waits_for_complete_training_prefix"}],"manualChecks":["Inspect the lagged-response example's leading unavailable values, coefficient parameters, and first usable derived sample."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","criteria":[{"criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"covered","evidence":["src/anomalyzer/relationships.py:115","tests/test_agentic.py:254"],"notes":"Supplied lag/coefficient uses the declared earlier predictor position and preserves the expected response residual."},{"criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"covered","evidence":["src/anomalyzer/relationships.py:152","tests/test_agentic.py:254","examples/agentic/lagged-response.json"],"notes":"Available and pending lineage retains existing source indexes and timestamp cutoffs; missing predictors do not become zero."},{"criterion":"When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","status":"covered","evidence":["src/anomalyzer/relationships.py:47","tests/test_agentic.py:294","tests/test_agentic.py:330"],"notes":"The coefficient is fitted once after the complete declared prefix; the independent 20/7 result and prefix-invariance assertions pass."},{"criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"covered","evidence":["src/anomalyzer/relationships.py:122","tests/test_agentic.py:330"],"notes":"Short prefixes retain null derived values and explicit unavailable reasons, while an unidentifiable completed fit has a visible error."}],"ignore_findings":[],"done":true,"input_fingerprint":"fc4516ef6ab903dfa4918268d97049aff04fbf9556620ccb8259ff830ba85785","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Check delayed responses"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878},{"id":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","label":"Inspect related metrics be...","area":"relationships","scopes":["relationships","relationships/resolution"],"scope":"relationships/resolution","scopeLabel":"relationships/resolution","status":"implemented","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","acceptance":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"guidance":["Daily ratios and differences are supported. Daily lagged, normalized-residual and joint-condition definitions remain explicitly inapplicable; their native evidence remains eligible. Local-calendar aggregation is separate work.","Initial support should use explicit relationships and compatible boundaries; cross-cadence joining or automatic resampling needs a separate contract.","Proposed automated checks cover ratio-of-totals arithmetic, missing sources, mismatched period boundaries, and open-period evidence separation.","Manual review should inspect a conversion decline within an unfinished day and verify the explanation separates native evidence from partial totals."],"path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","dependencies":["data/quality/analyze-sparse-observations","relationships/comparisons/compare-declared-metrics"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","src/anomalyzer/relationships.py","src/anomalyzer/agent.py","tests/test_sparse.py","examples/sparse/unfinished_daily.json","docs/sparse-data.md"],"automatedChecks":[{"id":"ratio-of-sums","description":"Compute daily ratios from compatible sums and preserve missing-child coverage.","command":"python -m pytest -q tests/test_sparse.py::test_complete_daily_ratio_divides_sums_and_missing_child_invalidates_only_its_view"},{"id":"view-policies","description":"Keep native/daily rules separate and preserve case identity on replay.","command":"python -m pytest -q tests/test_sparse.py::test_daily_rules_are_not_silently_reused_from_hourly_rules tests/test_sparse.py::test_multi_resolution_mcp_drilldown_selects_correct_view_and_replays_case_ids"},{"id":"view-limits","description":"Report unsupported daily lag semantics without discarding native evidence.","command":"python -m pytest -q tests/test_sparse.py::test_unsupported_daily_relationship_is_explicit_and_does_not_reinterpret_lag"}],"manualChecks":["Inspect progression chapter 22 and confirm incomplete progress is not presented as a completed total or independent corroboration."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for relationships/resolution/compare-finalized-subperiods.","criteria":[{"criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:22","src/anomalyzer/sparse_views.py:139","src/anomalyzer/sparse_views.py:143","src/anomalyzer/sparse_views.py:164","src/anomalyzer/agent.py:14","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:23","src/anomalyzer/agent.py:29","tests/test_sparse.py:18","tests/test_sparse.py:59","tests/test_sparse.py:87","tests/test_sparse.py:99","tests/test_sparse.py:117","docs/sparse-data.md:31","docs/sparse-data.md:127","docs/sparse-data.md:159","docs/sparse-data.md:199","docs/sparse-data.md:203"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:122","src/anomalyzer/sparse_views.py:123","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:36","src/anomalyzer/agent.py:34","tests/test_sparse.py:10","tests/test_sparse.py:40","tests/test_sparse.py:50","tests/test_sparse.py:60","tests/test_sparse.py:62","docs/sparse-data.md:17","docs/sparse-data.md:41","docs/sparse-data.md:68","docs/sparse-data.md:97","docs/sparse-data.md:99"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:70","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:170","src/anomalyzer/relationships.py:90","src/anomalyzer/relationships.py:173","tests/test_sparse.py:230","tests/test_sparse.py:259","tests/test_sparse.py:313","docs/sparse-data.md:41","docs/sparse-data.md:111","docs/sparse-data.md:194","docs/sparse-data.md:197","docs/sparse-data.md:198"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:130","src/anomalyzer/sparse_views.py:159","src/anomalyzer/relationships.py:157","src/anomalyzer/relationships.py:192","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:47","tests/test_sparse.py:217","tests/test_sparse.py:247","tests/test_sparse.py:251","tests/test_sparse.py:264","tests/test_sparse.py:386","examples/sparse/unfinished_daily.json:208","docs/sparse-data.md:8","docs/sparse-data.md:11","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:45"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","partial: Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","partial: Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","partial: Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","name":"Related-metric analysis","label":"Related-metric analysis / Monitor the periods my business uses / Compare related metrics while the period is open"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878}],"links":[{"source":"data/inputs/accept-analysis-requests","target":"analysis/evidence/calibrate-departures","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/cli/analyze-files-and-streams","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"analysis/expectations/model-seasonal-level-and-growth","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"interfaces/python/analyze-through-public-api","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/identify-residual-patterns","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"investigation/assessments/describe-early-changes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"relationships/quality/require-compatible-sources","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/expectations/reset-declared-regimes","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"data/quality/exclude-incomplete-periods","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/mcp/connect-read-only-local-server","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/rules/check-declared-numerical-rules","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/report-point-anomaly-episodes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/evidence/assess-reference-readiness","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/adapt-to-changing-growth","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/protect-reference-from-extremes","kind":"committed"},{"source":"data/quality/exclude-incomplete-periods","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"execution/resilience/retain-evidence-under-failure-and-limits","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/mcp/connect-read-only-local-server","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/visualize-analysis-on-request","kind":"committed"},{"source":"investigation/rules/check-declared-numerical-rules","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/evaluate-delayed-response","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"},{"source":"relationships/comparisons/compare-declared-metrics","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"}]};
+const graph = {"nodes":[{"id":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","label":"Bring my metrics into an a...","area":"data","scopes":["data","data/inputs"],"scope":"data/inputs","scopeLabel":"data/inputs","status":"implemented","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","acceptance":["Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","Published schema 1.0 and 1.1 request and result schemas match the public Python contracts."],"guidance":["Null observations remain unknown unless a schema-1.1 additive-volume policy explicitly authorizes confirmed empty collection as zero. Limits apply to the request as a whole.","Describe current supported behavior; historical plans do not add accepted inputs."],"path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["analysis/evidence/calibrate-departures","analysis/expectations/model-seasonal-level-and-growth","data/quality/analyze-sparse-observations","data/quality/identify-distinct-events","execution/resilience/retain-evidence-under-failure-and-limits","interfaces/python/analyze-through-public-api","investigation/assessments/describe-early-changes"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_agentic.py","schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json"],"automatedChecks":[{"id":"input-forms","description":"Analyze a bare list and validate a canonical request and result.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings tests/test_analysis.py::test_constant_and_schema"},{"id":"explicit-scope","description":"Reject schema 1.1 datasets without entity scope.","command":"python -m pytest -q tests/test_agentic.py::test_schema_11_requires_explicit_entity_scope"},{"id":"invalid-inputs","description":"Reject unsupported settings and nonnumeric or nonfinite observations.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_config tests/test_analysis.py::test_invalid_values"},{"id":"published-schemas","description":"Compare all four checked-in schemas with their public contracts.","command":"python -m pytest -q tests/test_analysis.py::test_checked_in_schemas tests/test_agentic.py::test_v11_checked_in_schemas_exist"}],"manualChecks":["Review the input examples in README.md and confirm they describe the two supported schema versions and the four-dataset/four-relationship limit."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","criteria":[{"criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"covered","evidence":["src/anomalyzer/recipes.py:21","src/anomalyzer/contracts.py:96","tests/test_analysis.py:234"],"notes":"Bare-list settings are validated; schema 1.0 limits datasets to exactly one."},{"criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"covered","evidence":["src/anomalyzer/contracts.py:240","src/anomalyzer/contracts.py:335","tests/test_agentic.py:192"],"notes":"Entity scope is required, relationship kinds are discriminated, and references must name distinct known datasets."},{"criterion":"Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:7","src/anomalyzer/contracts.py:71","tests/test_analysis.py:526","tests/test_analysis.py:531"],"notes":"Assertions reject each listed category; missing null values remain a separate accepted representation."},{"criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"covered","evidence":["schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json","tests/test_analysis.py:547","tests/test_agentic.py:317"],"notes":"Tests and independent inspection compare each complete published schema with its contract-generated schema. Python cross-field validators supplement JSON Schema constraints."}],"ignore_findings":[],"done":true,"input_fingerprint":"3546cd8f436812ff11c9352a855926677b55cc6287e40f6a5041bd613173e5b4","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Supply a numeric series"},"impact":7,"gaps":0,"r":73,"x":353.33333333333337,"y":170},{"id":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","label":"Understand how much histor...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","acceptance":["Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","Provisional residual scores can be reported but do not trigger point-anomaly episodes.","Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger."],"guidance":["Evidence strength is not a probability of an incident.","Calibration and expectations are separate concerns; each declared regime calibrates again."],"path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/detection/identify-residual-patterns","analysis/detection/report-point-anomaly-episodes","analysis/evidence/assess-reference-readiness","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","relationships/comparisons/compare-declared-metrics"],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"short-history","description":"Keep reference-only evidence for a short series.","command":"python -m pytest -q tests/test_analysis.py::test_short_series_returns_reference_evidence"},{"id":"provisional-evidence","description":"Report provisional scores without point triggers.","command":"python -m pytest -q tests/test_analysis.py::test_early_evidence_maturity_and_no_provisional_trigger"},{"id":"independent-calibration","description":"Check a known residual center, sample scale, and standardized score.","command":"python -m pytest -q tests/test_analysis.py::test_calibration_reference"},{"id":"constant-reference","description":"Handle zero residual variance without spurious episodes.","command":"python -m pytest -q tests/test_analysis.py::test_constant_and_schema"}],"manualChecks":["Inspect a sample's signal_maturity, calibration_samples, standardized_residual, and triggers together before describing a statistical departure."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","criteria":[{"criterion":"Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","status":"covered","evidence":["src/anomalyzer/recipes.py:205","src/anomalyzer/recipes.py:445","tests/test_analysis.py:141"],"notes":"The short-series test checks three reference-only samples, absent standardized scores, and no triggers."},{"criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"covered","evidence":["src/anomalyzer/recipes.py:210","tests/test_analysis.py:123"],"notes":"A provisional score of 2 exceeds the configured point threshold of 1 but produces no trigger or episode."},{"criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"covered","evidence":["src/anomalyzer/recipes.py:231","src/anomalyzer/recipes.py:307","tests/test_analysis.py:111"],"notes":"Calibration uses retained residual mean and sample standard deviation once per regime; the independent numerical example produces the expected standardized score."},{"criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"covered","evidence":["src/anomalyzer/contracts.py:104","src/anomalyzer/recipes.py:350","src/anomalyzer/recipes.py:388","tests/test_analysis.py:111"],"notes":"Sample fields retain the comparison and support; the point threshold is inspectable in the enclosing method parameters."}],"ignore_findings":[],"done":true,"input_fingerprint":"ccdc1b2e92f3e647d5a9e6127cc3c4b10bf0da8f1f78af3efb805daa9b1e4305","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Establish evidence support"},"impact":6,"gaps":0,"r":68,"x":170,"y":170},{"id":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","label":"Analyze local files from t...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","acceptance":["The methods command lists supported analysis recipes and residual detectors.","Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status."],"guidance":["CSV callers must select one entity and unit before analysis.","Exit 0 includes completed analysis with anomalies, insufficient history, or inapplicability; exit 2 means invalid input, and exit 1 means execution failure or partial execution."],"path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/python/install-versioned-library","interfaces/skill/guide-anomaly-investigation","interfaces/skill/visualize-analysis-on-request"],"implementationReferences":["src/anomalyzer/cli.py","src/anomalyzer/io.py","src/anomalyzer/__main__.py","tests/test_cli.py"],"automatedChecks":[{"id":"file-stdin","description":"Analyze equivalent file and stdin requests.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence"},{"id":"csv-precedence","description":"Validate selected CSV fields and apply flag-over-file-over-embedded precedence.","command":"python -m pytest -q tests/test_cli.py::test_csv_and_precedence"},{"id":"value-list","description":"Analyze a bare JSON array and expose inferred settings.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"},{"id":"output-and-exits","description":"Protect existing output and distinguish invalid options from failed execution.","command":"python -m pytest -q tests/test_cli.py::test_output_protection tests/test_cli.py::test_methods_and_invalid_options tests/test_cli.py::test_size_limit_and_execution_exit"}],"manualChecks":["Run an example in readable and JSON modes and confirm stdout contains results while diagnostics go to stderr."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","criteria":[{"criterion":"The methods command lists supported analysis recipes and residual detectors.","status":"covered","evidence":["src/anomalyzer/cli.py:426","src/anomalyzer/registry.py:15","tests/test_cli.py:27"],"notes":"Discovery tests check all three recipes and ten residual detectors, with structured stdout and no successful-run diagnostics on stderr."},{"criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:239","tests/test_cli.py:13","tests/test_mcp.py:163"],"notes":"The passing file/stdin regression compares request fingerprints and short-history semantics; cross-interface tests compare complete stable numerical output."},{"criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"covered","evidence":["src/anomalyzer/io.py:33","tests/test_cli.py:51"],"notes":"The passing CSV regression covers timestamped missing cells, positional data, and varying unmapped entity rejection."},{"criterion":"Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","status":"covered","evidence":["src/anomalyzer/cli.py:253","tests/test_cli.py:51","tests/test_cli.py:71"],"notes":"The passing file/flag/embedded precedence regression and bare-values regression establish the merge order and validated configuration."},{"criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"covered","evidence":["src/anomalyzer/io.py:69","src/anomalyzer/cli.py:424","tests/test_cli.py:41","tests/test_cli.py:108"],"notes":"Output protection and exit regressions passed, including explicit overwrite; successful manual examples have empty stderr and incomplete-usage exits 1."}],"ignore_findings":[],"done":true,"input_fingerprint":"86e701ccc0a1aefd407f11b4547784fef497fa1b49ddf6bb69db0d08f757dfdc","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Run from files or stdin"},"impact":4,"gaps":0,"r":58,"x":903.3333333333334,"y":170},{"id":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","label":"Spot sustained shifts and ...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","acceptance":["Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","Trend, oscillation, and mixture findings retain their specific pattern meanings.","Moving-range findings identify variation changes separately from location changes."],"guidance":["Pattern findings may overlap and are not independent confirmations.","A residual location shift does not establish a change in the full distribution."],"path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations","data/quality/identify-distinct-events","investigation/cases/group-related-evidence"],"implementationReferences":["src/anomalyzer/evidence.py","src/anomalyzer/recipes.py","tests/test_analysis.py"],"automatedChecks":[{"id":"location-shift","description":"Find moderate increases and decreases without point anomalies.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_patterns_describe_location_shift_without_point_anomaly"},{"id":"accumulating-change","description":"Detect a known CUSUM accumulation at the expected position.","command":"python -m pytest -q tests/test_analysis.py::test_cusum_detects_accumulated_moderate_location_shift"},{"id":"diagnostic-patterns","description":"Distinguish residual trend, oscillation, and mixture rules.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_diagnostics_have_specific_semantics"},{"id":"variation","description":"Describe a moving-range increase as variation rather than location.","command":"python -m pytest -q tests/test_analysis.py::test_moving_range_detects_short_term_variation_increase"}],"manualChecks":["Compare examples/location_shift_up.json with its point episodes and pattern findings, allowing for overlapping residual rules."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","criteria":[{"criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"covered","evidence":["src/anomalyzer/evidence.py:118","src/anomalyzer/evidence.py:189","tests/test_analysis.py:293"],"notes":"Tests assert no point triggers or episodes and exact upward/downward detection positions for Nelson rules and CUSUM."},{"criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"covered","evidence":["src/anomalyzer/evidence.py:211","src/anomalyzer/evidence.py:230","tests/test_analysis.py:350"],"notes":"The known accumulation detects at index 16 and reports peak CUSUM approximately 6, threshold 5, and increase direction."},{"criterion":"Trend, oscillation, and mixture findings retain their specific pattern meanings.","status":"covered","evidence":["src/anomalyzer/evidence.py:70","src/anomalyzer/evidence.py:122","tests/test_analysis.py:340"],"notes":"Separate predicates and descriptions preserve the three meanings and explicitly limit location/distribution claims."},{"criterion":"Moving-range findings identify variation changes separately from location changes.","status":"covered","evidence":["src/anomalyzer/evidence.py:234","tests/test_analysis.py:360"],"notes":"Adjacent standardized-residual differences produce variation_shift findings; the test checks magnitude, detection index, and interpretation."}],"ignore_findings":[],"done":true,"input_fingerprint":"63716eb67db6b73964b79897d8c81f2891df2586ff213c95ea81c5699b8abbea","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Find sustained residual changes"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","label":"Account for ordinary seaso...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","acceptance":["A supplied season length selects earlier matching seasonal observations for the reference.","Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected."],"guidance":["Expectations support scoring; they are not a standalone forecasting API."],"path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/expectations/protect-reference-from-extremes","analysis/expectations/reset-declared-regimes","analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py"],"automatedChecks":[{"id":"seasonal-reference","description":"Check independent lagged-reference values.","command":"python -m pytest -q tests/test_analysis.py::test_forecast_independent_reference"},{"id":"growth-and-departures","description":"Keep clean linear and compound growth ordinary while identifying injected departures.","command":"python -m pytest -q tests/test_trend.py::test_trend_and_anomalies"},{"id":"frozen-fit","description":"Keep the training fit stable when later observations change.","command":"python -m pytest -q tests/test_trend.py::test_frozen_fit_and_legacy_mode"},{"id":"explicit-trends","description":"Accept supported trend choices and handle short, constant, and signed series.","command":"python -m pytest -q tests/test_trend.py::test_explicit_modes_and_validation tests/test_trend.py::test_short_history_constant_and_signed_linear"}],"manualChecks":["Compare examples/growth_up.json and examples/compound.json with their selected trend and evidence expectations."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","criteria":[{"criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"covered","evidence":["src/anomalyzer/baseline.py:8","src/anomalyzer/baseline.py:31","tests/test_analysis.py:21"],"notes":"Independent lag values and robust matching-phase assertions establish the seasonal reference calculation."},{"criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"covered","evidence":["src/anomalyzer/trend.py:120","src/anomalyzer/trend.py:144","tests/test_trend.py:11"],"notes":"Positive and negative linear/exponential cases select the expected model, match post-training expectations, and produce no ordinary-growth episodes."},{"criterion":"The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","status":"covered","evidence":["src/anomalyzer/recipes.py:162","tests/test_trend.py:36"],"notes":"Trend fitting occurs at the training boundary; the test asserts identical trend diagnostics and earlier evidence after later observations change."},{"criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:20","tests/test_trend.py:49","tests/test_trend.py:57"],"notes":"Tests exercise disabled, linear, exponential, and invalid quadratic modes, including disclosed short-history fallback."}],"ignore_findings":[],"done":true,"input_fingerprint":"8e9a0361b02b7d3f651366e5fa0675cccb3949e44de448d2ae0d43e6710bd408","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Account for growth and seasonality"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","label":"Embed local analysis in my...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","acceptance":["analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","analyze_series preserves supplied settings and event-time context for a single dataset.","analyze_relationships preserves the declared multi-dataset relationship semantics.","Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime."],"guidance":["All interfaces use local analysis; there is no hosted service or model API requirement."],"path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence","interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/__init__.py","src/anomalyzer/agent.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"request-parity","description":"Match canonical request results across Python, CLI, and full-result MCP compatibility calls.","command":"python -m pytest -q tests/test_mcp.py::test_canonical_request_matches_python_cli_and_mcp tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"series-context","description":"Preserve a single-series event-time cutoff and context across interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp tests/test_mcp.py::test_series_wrapper_preserves_omitted_and_explicit_season_settings"},{"id":"relationship-parity","description":"Match declared relationship evidence across the public interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_relationship_semantics_match_python_cli_and_mcp"}],"manualChecks":["Review the Python API examples in README.md and confirm callers can inspect the returned validated models and JSON."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","criteria":[{"criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"covered","evidence":["src/anomalyzer/__init__.py:3","src/anomalyzer/recipes.py:21","tests/test_analysis.py:234"],"notes":"Bare values and separate settings validate before dispatch; canonical schema 1.0 and 1.1 requests use their respective contracts."},{"criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"covered","evidence":["src/anomalyzer/agent.py:11","tests/test_mcp.py:191","tests/test_mcp.py:291"],"notes":"Field-set handling preserves omitted versus explicit settings and context in the single-series wrapper, verified across dictionary and Settings inputs."},{"criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"covered","evidence":["src/anomalyzer/agent.py:27","tests/test_mcp.py:217"],"notes":"Declared relationships validate and use the shared engine without changing units, lineage, or primitive semantics."},{"criterion":"Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime.","status":"covered","evidence":["tests/test_mcp.py:163","tests/test_mcp.py:191","tests/test_mcp.py:217","src/anomalyzer/__main__.py:1"],"notes":"Passing real CLI subprocess, Python, and full-result MCP parity tests compare stable numerical results apart from identity and runtime."}],"ignore_findings":[],"done":true,"input_fingerprint":"214bd10bb3aa5fad0f74938fe2ef09853a593cdb5b93cba49ec851a628d8ab65","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","name":"Python integration","label":"Python integration / Embed analysis in my application / Call the public API"},"impact":3,"gaps":0,"r":53,"x":903.3333333333334,"y":170},{"id":"investigation/cases/group-related-evidence","title":"Investigate related findings together","label":"Investigate related findin...","area":"investigation","scopes":["investigation","investigation/cases"],"scope":"investigation/cases","scopeLabel":"investigation/cases","status":"implemented","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","acceptance":["Evidence for different declared entities remains in separate cases even at the same event time.","Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","A derived relationship adds context without counting as an additional independent source for severity.","Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection."],"guidance":["Warning and critical are deterministic numerical grouping rules, not validated incident impact.","Diagnostic trend, oscillation, mixture, and variation findings remain distinct from adaptive location-shift cases."],"path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","dependencies":["investigation/assessments/describe-early-changes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/mcp/investigate-findings-and-evidence","investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/cases.py","src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py"],"automatedChecks":[{"id":"entity-scope","description":"Separate cases for different entities at the same event time.","command":"python -m pytest -q tests/test_agentic.py::test_cases_are_scoped_by_entity_at_the_same_event_time"},{"id":"source-severity","description":"Escalate numerical severity when two source datasets meet criteria.","command":"python -m pytest -q tests/test_agentic.py::test_two_source_datasets_escalate_case_severity"},{"id":"correlated-context","description":"Keep relationship evidence from double-counting source severity.","command":"python -m pytest -q tests/test_agentic.py::test_relationship_evidence_explains_but_does_not_double_count_severity"},{"id":"first-detection","description":"Group an adaptive location shift before point triggers without using a later pattern peak.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_location_shift_creates_one_early_case_without_point_trigger tests/test_agentic.py::test_location_case_severity_does_not_use_future_pattern_peak"}],"manualChecks":["Inspect a case's contributing targets, explanation, evidence references, and investigation questions; distinguish it from a point-anomaly episode."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","criteria":[{"criterion":"Evidence for different declared entities remains in separate cases even at the same event time.","status":"covered","evidence":["src/anomalyzer/cases.py:42","tests/test_agentic.py:69"],"notes":"Entity participates in grouping and stable identity; the test requires separate cases and distinct IDs at the same time."},{"criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"covered","evidence":["src/anomalyzer/cases.py:72","tests/test_agentic.py:85"],"notes":"Two qualifying source datasets produce critical numerical severity and named source explanations."},{"criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"covered","evidence":["src/anomalyzer/cases.py:72","src/anomalyzer/cases.py:116","tests/test_agentic.py:100"],"notes":"The independent-source count includes dataset criteria only. Derived evidence remains explanatory and can still meet the separately declared magnitude rule."},{"criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"covered","evidence":["src/anomalyzer/early.py:193","src/anomalyzer/early.py:237","src/anomalyzer/orchestrator.py:115","tests/test_agentic.py:122","tests/test_agentic.py:146"],"notes":"Tests require one moderate-shift case without point triggers and restrict assessment references/strength to samples at or before first detection."}],"ignore_findings":[],"done":true,"input_fingerprint":"6830af8a2f064123c12adb062f4793e446b97b3b805fd73a934ea84cd0d5d9cf","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Investigate related evidence"},"impact":3,"gaps":0,"r":53,"x":1086.6666666666667,"y":170},{"id":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","label":"Assess sparse observations...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","acceptance":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"guidance":["Default missingness is unknown. A modeled expectation is never a measurement.","Caller business severity, incident state and notifications remain outside the engine.","Optional event-identity-v1 IDs retain simultaneous distinct calls; only a separately asserted trusted sequence permits ordering their measurements."],"path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["data/quality/identify-distinct-events","relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","src/anomalyzer/orchestrator.py","src/anomalyzer/relationships.py","src/anomalyzer/early.py","tests/test_sparse.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"source-semantics","description":"Preserve unknown periods and require explicit empty-volume authorization.","command":"python -m pytest -q tests/test_sparse.py -k \"zero or omitted or grid or duplicate or expansion\""},{"id":"causal-boundaries","description":"Retain real phase, event times, resets and as-of revision eligibility.","command":"python -m pytest -q tests/test_sparse.py -k \"phase or gap or reset or event or revision or arrival\""},{"id":"public-interfaces","description":"Preserve useful rules, relationships, views, cases and API parity.","command":"python -m pytest -q tests/test_sparse.py -k \"rule or relationship or daily or parity\""},{"id":"progressive-examples","description":"Reproduce the full numbered sequence and sparse worked examples.","command":"python examples/run_demo.py"}],"manualChecks":["Inspect progression figures 14–22 for true gaps, distinguished policy zeros, original event time, unavailable checks and caller-owned business severity."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/analyze-sparse-observations.","criteria":[{"criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","status":"partial","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:71","src/anomalyzer/contracts.py:85","src/anomalyzer/contracts.py:86","src/anomalyzer/sparse.py:3","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:42","src/anomalyzer/sparse.py:48","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse_views.py:74","src/anomalyzer/sparse_views.py:81","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:97","src/anomalyzer/sparse_views.py:119","src/anomalyzer/orchestrator.py:63","src/anomalyzer/orchestrator.py:127","src/anomalyzer/orchestrator.py:139","src/anomalyzer/orchestrator.py:191","src/anomalyzer/relationships.py:36","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:98","src/anomalyzer/relationships.py:131","src/anomalyzer/relationships.py:141","src/anomalyzer/early.py:58","src/anomalyzer/early.py:137","tests/test_sparse.py:44","tests/test_sparse.py:45","tests/test_sparse.py:50","tests/test_sparse.py:51","tests/test_sparse.py:57","docs/sparse-data.md:4","docs/sparse-data.md:29","docs/sparse-data.md:37","docs/sparse-data.md:39","docs/sparse-data.md:40","examples/sparse_examples.py:22","examples/sparse_examples.py:23","examples/sparse_examples.py:25","examples/sparse_examples.py:26","examples/sparse_examples.py:28","examples/render_readme_progression.py:135","examples/render_readme_progression.py:139","examples/render_readme_progression.py:140","examples/render_readme_progression.py:166","examples/render_readme_progression.py:206"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","status":"partial","evidence":["src/anomalyzer/contracts.py:27","src/anomalyzer/contracts.py:49","src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:56","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:181","src/anomalyzer/sparse.py:187","src/anomalyzer/sparse.py:189","src/anomalyzer/sparse_views.py:15","src/anomalyzer/sparse_views.py:98","src/anomalyzer/sparse_views.py:107","src/anomalyzer/sparse_views.py:108","src/anomalyzer/sparse_views.py:112","src/anomalyzer/orchestrator.py:22","src/anomalyzer/orchestrator.py:23","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:129","src/anomalyzer/orchestrator.py:141","src/anomalyzer/relationships.py:185","src/anomalyzer/relationships.py:189","src/anomalyzer/relationships.py:191","src/anomalyzer/early.py:46","src/anomalyzer/early.py:48","src/anomalyzer/early.py:51","src/anomalyzer/early.py:59","src/anomalyzer/early.py:61","tests/test_sparse.py:20","tests/test_sparse.py:108","tests/test_sparse.py:111","tests/test_sparse.py:113","tests/test_sparse.py:119","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:42","docs/sparse-data.md:106","docs/sparse-data.md:108","examples/sparse_examples.py:16","examples/sparse_examples.py:33","examples/sparse_examples.py:50","examples/sparse_examples.py:51","examples/sparse_examples.py:91","examples/render_readme_progression.py:241","examples/render_readme_progression.py:312","examples/render_readme_progression.py:317","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:242","src/anomalyzer/contracts.py:245","src/anomalyzer/contracts.py:258","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:57","src/anomalyzer/sparse.py:63","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:127","src/anomalyzer/sparse_views.py:155","src/anomalyzer/sparse_views.py:157","src/anomalyzer/orchestrator.py:37","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:178","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:40","src/anomalyzer/relationships.py:57","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:96","src/anomalyzer/relationships.py:97","src/anomalyzer/early.py:140","src/anomalyzer/early.py:147","tests/test_sparse.py:59","tests/test_sparse.py:67","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:128","docs/sparse-data.md:26","docs/sparse-data.md:96","docs/sparse-data.md:101","docs/sparse-data.md:132","docs/sparse-data.md:142","examples/sparse_examples.py:26","examples/sparse_examples.py:36","examples/sparse_examples.py:38","examples/sparse_examples.py:57","examples/sparse_examples.py:81","examples/render_readme_progression.py:654","examples/render_readme_progression.py:697","examples/render_readme_progression.py:884","examples/render_readme_progression.py:896","examples/render_readme_progression.py:917"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","status":"partial","evidence":["src/anomalyzer/contracts.py:13","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:297","src/anomalyzer/contracts.py:298","src/anomalyzer/contracts.py:437","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse.py:66","src/anomalyzer/sparse.py:83","src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:28","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:94","src/anomalyzer/orchestrator.py:106","src/anomalyzer/orchestrator.py:107","src/anomalyzer/orchestrator.py:133","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:160","src/anomalyzer/relationships.py:162","src/anomalyzer/relationships.py:166","tests/test_sparse.py:59","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:123","tests/test_sparse.py:180","docs/sparse-data.md:26","docs/sparse-data.md:41","docs/sparse-data.md:69","docs/sparse-data.md:78","docs/sparse-data.md:96","examples/sparse_examples.py:59","examples/sparse_examples.py:95","examples/render_readme_progression.py:686","examples/render_readme_progression.py:699","examples/render_readme_progression.py:710","examples/render_readme_progression.py:767","examples/render_readme_progression.py:884"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","partial: Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","partial: Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","partial: As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Assess sparse source data"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","label":"Analyze observations in th...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","acceptance":["Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","Calendar duplicate timestamps require an aggregation policy; identified event-index calls can share original timestamps while duplicate logical IDs are rejected and arbitrary tie order supplies no sequence evidence.","An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration."],"guidance":["No interpolation or implicit filling, resampling, or unit conversion occurs. Schema-1.1 source policies may explicitly authorize confirmed empty additive volume as zero.","Duplicate null values propagate through aggregation.","Stable event identity and trusted sequence are separate caller contracts; legacy event inputs without IDs still require unique timestamps."],"path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["data/quality/exclude-incomplete-periods","relationships/quality/require-compatible-sources"],"implementationReferences":["src/anomalyzer/events.py","tests/test_events.py","docs/sparse-data.md","src/anomalyzer/sparse.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/contracts.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"coordinate-contract","description":"Reject partial timestamp metadata and positional as_of cutoffs.","command":"python -m pytest -q tests/test_analysis.py::test_timestamp_metadata_is_all_or_nothing"},{"id":"local-time","description":"Normalize valid local times and reject ambiguous or nonexistent times.","command":"python -m pytest -q tests/test_analysis.py::test_local_time"},{"id":"duplicates-cutoff","description":"Require an explicit duplicate policy and retain counts after the cutoff.","command":"python -m pytest -q tests/test_analysis.py::test_duplicate_sort_and_cutoff tests/test_events.py::test_distinct_same_time_calls_keep_values_references_and_cases"},{"id":"cadence-context","description":"Check fixed weekly cadence and preserve cutoff semantics across interfaces.","command":"python -m pytest -q tests/test_analysis.py::test_weekly_cadence tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp"}],"manualChecks":["Inspect a timestamped result's data_quality transformations, UTC window, and resolved context to confirm preparation is explainable."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","criteria":[{"criterion":"Timestamped datasets provide both timestamps and frequency; positional datasets provide neither.","status":"covered","evidence":["src/anomalyzer/contracts.py:77","tests/test_analysis.py:418"],"notes":"Partial coordinate metadata and positional as_of are rejected."},{"criterion":"Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","status":"covered","evidence":["src/anomalyzer/validation.py:8","tests/test_analysis.py:500"],"notes":"UTC round trips detect ambiguous/nonexistent local times; valid local time normalizes to the expected UTC hour."},{"criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"covered","evidence":["src/anomalyzer/validation.py:87","src/anomalyzer/validation.py:103","tests/test_analysis.py:478"],"notes":"Explicit mean/sum handling retains duplicate counts and transformation descriptions. Nulls propagate rather than becoming zero."},{"criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"covered","evidence":["src/anomalyzer/validation.py:105","src/anomalyzer/validation.py:111","tests/test_analysis.py:491","tests/test_mcp.py:191"],"notes":"Filtering is inclusive through as_of; cadence compares UTC elapsed timedeltas, including fixed seven-day weeks."}],"ignore_findings":[],"done":true,"input_fingerprint":"791748764cdc0dfe81d987a908179930d858670606e51b64407172295776b261","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Prepare the comparison"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","label":"Let an agent inspect the e...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","acceptance":["The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","get_evidence exposes the selected finding's numerical support and relationship source lineage.","Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated."],"guidance":["Run results are retained only in the bounded server-session cache; missing or evicted run IDs require reanalysis.","Cases, episodes, and residual patterns may overlap."],"path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":["interfaces/mcp/connect-read-only-local-server","interfaces/skill/guide-anomaly-investigation"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py"],"automatedChecks":[{"id":"case-drilldown","description":"Analyze, page findings, and inspect exact conversion case values and lineage.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case tests/test_mcp.py::test_schema_11_episode_is_paged_and_has_relationship_evidence"},{"id":"unavailable-finding","description":"Expose a quality finding when latest relationship evidence is unavailable.","command":"python -m pytest -q tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"},{"id":"early-candidate","description":"Inspect descriptive candidate evidence without inventing a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case tests/test_mcp.py::test_sparse_insufficient_history_keeps_descriptive_candidate_visible"},{"id":"sample-pagination","description":"Page supporting samples for a residual pattern.","command":"python -m pytest -q tests/test_mcp.py::test_get_evidence_pages_pattern_samples"}],"manualChecks":["Follow analyze, list_findings, and get_evidence for the conversion example and check that the headline is supported by the retrieved values."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","criteria":[{"criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"covered","evidence":["src/anomalyzer/mcp.py:170","src/anomalyzer/mcp_views.py:247","tests/test_mcp.py:54"],"notes":"The conversion manual call returns a relationship headline, preview, counts, readiness, and session-scoped run ID."},{"criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:132","src/anomalyzer/mcp_views.py:260","tests/test_mcp.py:313","tests/test_mcp.py:108"],"notes":"Cases, early candidates, episodes, patterns, and quality findings feed a bounded page; relationship episode regression verifies schema 1.1 pagination."},{"criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:273","src/anomalyzer/mcp_views.py:294","tests/test_mcp.py:54","tests/test_mcp.py:313"],"notes":"Manual drilldown returns observed 8/130 versus expected 0.1 and source indexes orders=14 and qualified_visits=14."},{"criterion":"Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:118","src/anomalyzer/mcp_views.py:169","src/anomalyzer/mcp_views.py:337","tests/test_mcp.py:87","tests/test_mcp.py:137","tests/test_mcp.py:341"],"notes":"Passing tests retain short-history and inapplicable-method candidates plus unavailable latest evidence, while pattern samples are paginated."}],"ignore_findings":[],"done":true,"input_fingerprint":"a09b0003780bcd2f9ccf8e3d8ff58b3147c118760b984757d7027b5810fd9bef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Drill into findings and evidence"},"impact":2,"gaps":0,"r":48,"x":927.3333333333334,"y":288},{"id":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","label":"Spot early changes with a ...","area":"investigation","scopes":["investigation","investigation/assessments"],"scope":"investigation/assessments","scopeLabel":"investigation/assessments","status":"implemented","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","acceptance":["The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","Missing values break the comparison history, and zero previous values do not create relative-change evidence.","Early candidates remain non-triggering and ineligible for action even under a permissive policy."],"guidance":["An early candidate does not establish an anomaly, stable baseline, cause, or incident."],"path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["investigation/cases/group-related-evidence","investigation/rules/check-declared-numerical-rules"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_mcp.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"prefix-invariant-facts","description":"Describe short history causally without creating cases.","command":"python -m pytest -q tests/test_agentic.py::test_short_history_assessments_are_useful_and_prefix_invariant"},{"id":"comparison-boundaries","description":"Reset comparisons at missing observations and avoid dividing by zero.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"candidate-eligibility","description":"Keep candidates ineligible for action and notifications under permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"},{"id":"agent-visibility","description":"Expose a short-history candidate and its claim limits without a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case"}],"manualChecks":["Inspect a candidate's comparison population, maturity, established claim, and not_established text before describing it."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","criteria":[{"criterion":"The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","status":"covered","evidence":["src/anomalyzer/early.py:73","src/anomalyzer/early.py:96","tests/test_agentic.py:35"],"notes":"Assertions establish observation/change/change/departure_candidate for [10,12,11,20]."},{"criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"covered","evidence":["src/anomalyzer/early.py:50","src/anomalyzer/early.py:141","tests/test_agentic.py:35"],"notes":"Each assessment precedes the prior-history update; tests compare complete serialized assessments for every prefix."},{"criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"covered","evidence":["src/anomalyzer/early.py:53","src/anomalyzer/early.py:58","tests/test_capability_behaviors.py:87"],"notes":"The zero comparison has no relative change; after a missing value, comparison samples reset to zero."},{"criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"covered","evidence":["src/anomalyzer/early.py:105","src/anomalyzer/cases.py:35","tests/test_capability_behaviors.py:134","tests/test_mcp.py:341"],"notes":"Candidates have criterion_met=false and remain action/notification-ineligible after policy replay, without creating cases."}],"ignore_findings":[],"done":true,"input_fingerprint":"debdeeeaea559950f53461a4298cb9c2e3a1681f83a43e8a981ed9b51968a4d0","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect early evidence"},"impact":2,"gaps":0,"r":48,"x":1110.6666666666667,"y":288},{"id":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","label":"Compare the same entity an...","area":"relationships","scopes":["relationships","relationships/quality"],"scope":"relationships/quality","scopeLabel":"relationships/quality","status":"implemented","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","acceptance":["Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","Difference relationships require identical declared source units and matching output units.","An invalid relationship is reported as failed while valid sibling dataset results remain available."],"guidance":["A failed or unavailable relationship does not establish that its metric is normal.","No implicit timestamp matching, interpolation, unit conversion, or zero-filling occurs."],"path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"implementationReferences":["src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"exact-alignment","description":"Reject mismatched timestamps, cadence, entity, duplicates, and positional relationship sources.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relationship_requires_exact_source_alignment"},{"id":"compatible-units","description":"Report a difference unit mismatch without suppressing sibling results.","command":"python -m pytest -q tests/test_agentic.py::test_alignment_unit_and_partial_failure_are_specific"},{"id":"unambiguous-declaration","description":"Reject fields that do not belong to the selected relationship kind.","command":"python -m pytest -q tests/test_agentic.py::test_contract_rejects_ambiguous_relationship_fields"},{"id":"healthy-siblings","description":"Retain a healthy dataset's evidence when another dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Read both target-level statuses and errors before interpreting a partial run; check source dataset errors when a relationship source failed."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","criteria":[{"criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"covered","evidence":["src/anomalyzer/alignment.py:42","tests/test_capability_behaviors.py:30"],"notes":"Parameterized assertions require specific errors for timestamp, cadence, and entity mismatches."},{"criterion":"Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","status":"covered","evidence":["src/anomalyzer/alignment.py:23","src/anomalyzer/alignment.py:44","tests/test_capability_behaviors.py:30"],"notes":"Positional sources fail exact alignment; even an explicit duplicate-mean setting is rejected for relationship source lineage."},{"criterion":"Difference relationships require identical declared source units and matching output units.","status":"covered","evidence":["src/anomalyzer/relationships.py:41","tests/test_agentic.py:242"],"notes":"Source and output unit checks are explicit. Tests cover mismatched source units and successful matching units."},{"criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:171","tests/test_capability_behaviors.py:30","tests/test_capability_behaviors.py:159"],"notes":"Relationship errors remain target-specific, healthy dataset results survive, and the overall mixed run is partial."}],"ignore_findings":[],"done":true,"input_fingerprint":"06321ab5806a8276cc3954df5d6f4dd7f059f67600038854569787ff51b420d6","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Align comparable sources"},"impact":2,"gaps":0,"r":48,"x":1294,"y":288},{"id":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","label":"Compare against a new base...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","acceptance":["Integer positions and exact observed dates or timestamps can identify reset boundaries.","Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","Repeated, out-of-range, or unmatched reset points are rejected.","CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries."],"guidance":["Reset points are explicit caller decisions. Known-event notes and review-state fields do not rebuild regimes.","Multi-resolution analysis does not support reset_points."],"path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations"],"implementationReferences":["src/anomalyzer/validation.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py"],"automatedChecks":[{"id":"independent-regimes","description":"Retrain separate regimes without cross-boundary evidence or patterns.","command":"python -m pytest -q tests/test_analysis.py::test_manual_resets_retrain_without_crossing_boundaries"},{"id":"invalid-boundaries","description":"Reject invalid or duplicate resolved reset positions.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_manual_resets"},{"id":"cli-boundaries","description":"Resolve position and date reset flags through the CLI.","command":"python -m pytest -q tests/test_cli.py::test_manual_reset_cli"}],"manualChecks":["Inspect the resolved reset positions and segment diagnostics for examples/steps_reset.json."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","criteria":[{"criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"covered","evidence":["src/anomalyzer/validation.py:129","tests/test_analysis.py:381"],"notes":"Positions and observed dates are exercised end to end; timestamp normalization and exact matching are implemented directly."},{"criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"covered","evidence":["src/anomalyzer/recipes.py:119","src/anomalyzer/recipes.py:143","tests/test_analysis.py:381"],"notes":"Each segment initializes separate state and passes its own start to forecasts. Tests verify three calibrated regimes and no cross-boundary evidence or findings."},{"criterion":"Repeated, out-of-range, or unmatched reset points are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/validation.py:149","tests/test_analysis.py:413"],"notes":"Validation rejects literal duplicates and different declarations resolving to the same position, invalid positions, and unmatched dates/timestamps."},{"criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"covered","evidence":["tests/test_cli.py:82","src/anomalyzer/recipes.py:53"],"notes":"CLI assertions establish positions [50, 100] and removal of the unmarked-regime episodes."}],"ignore_findings":[],"done":true,"input_fingerprint":"067063f972c5b70d990d6cb0125892731a97a7f482698d7b331a1a30229ac223","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Restart after a reviewed change"},"impact":1,"gaps":0,"r":43,"x":194,"y":288},{"id":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","label":"Avoid mistaking unfinished...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","acceptance":["Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","The observed values and timestamps of unfinished periods remain available in data quality output.","Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","A relationship with an unavailable latest source reports that unavailability rather than normal behavior."],"guidance":["Completeness is declared by the caller; the engine does not infer collection completeness."],"path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/alignment.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"unfinished-values","description":"Retain an unfinished value without scoring it.","command":"python -m pytest -q tests/test_analysis.py::test_explicit_incomplete_period_is_reported_and_not_scored"},{"id":"trailing-only","description":"Reject incomplete observations followed by completed observations.","command":"python -m pytest -q tests/test_analysis.py::test_incomplete_periods_must_be_trailing"},{"id":"source-and-agent-views","description":"Exclude a partial source sample from both analysis views and expose unavailable relationship evidence.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"}],"manualChecks":["Inspect examples/agentic/incomplete-usage.json and confirm the latest headline explains missing completeness and does not imply normality."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicitly incomplete trailing observations are reported as data quality evidence and excluded from completed-series analysis.","criteria":[{"criterion":"Incomplete observations form a trailing suffix after chronological preparation.","status":"covered","evidence":["src/anomalyzer/validation.py:44","tests/test_analysis.py:162"],"notes":"Suffix validation runs after timestamp preparation and rejects a completed observation following an incomplete one."},{"criterion":"The observed values and timestamps of unfinished periods remain available in data quality output.","status":"covered","evidence":["src/anomalyzer/validation.py:50","src/anomalyzer/validation.py:114","tests/test_analysis.py:150"],"notes":"Quality records retain original source index, normalized timestamp, observed value, and incomplete status."},{"criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"covered","evidence":["src/anomalyzer/validation.py:53","src/anomalyzer/recipes.py:564","tests/test_analysis.py:150","tests/test_analysis.py:200"],"notes":"Only finalized prepared values enter either scoring path; assertions verify the last scored sample and both multi-resolution counts."},{"criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"covered","evidence":["src/anomalyzer/alignment.py:35","src/anomalyzer/orchestrator.py:150","tests/test_mcp.py:87"],"notes":"Incomplete aligned sources become unavailable values, and latest relationship state/status plus MCP headline explain the lack of evidence."}],"ignore_findings":[],"done":true,"input_fingerprint":"d82ee816fe89a25232e2589587d31bcadfa81be4188a0e1531f4fbfdf0ebcb5e","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Keep unfinished data visible"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","label":"Give an agent read-only ac...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","acceptance":["Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","Analysis tool calls return structured content together with compact text.","Invalid tool input and unknown session run IDs return structured errors rather than fabricated results."],"guidance":["Compatibility tool names remain callable but are not advertised.","Legacy clients receive isError tool results; modern clients receive JSON-RPC invalid-input errors.","The server provides analysis and evidence inspection; the caller owns monitoring and operational workflows."],"path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","dependencies":["interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":["interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py","pyproject.toml"],"automatedChecks":[{"id":"tool-discovery","description":"Advertise the three read-only tools and their schema contracts.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_lists_read_only_tools"},{"id":"connection-protocols","description":"Support legacy stdio initialization and modern stateless discovery.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_stdio_legacy_handshake_and_tool_listing tests/test_mcp.py::test_mcp_modern_discovery_and_stateless_tool_listing"},{"id":"structured-content","description":"Return a compact analysis response with structured content and run ID.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_tool_call_returns_structured_content"},{"id":"invalid-calls","description":"Reject malformed input and unknown session run IDs in both protocols.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_invalid_tool_input_is_a_tool_error tests/test_mcp.py::test_paged_tool_inputs_enforce_schema_in_both_protocols"}],"manualChecks":["Review discovery's interpretation guidance and confirm the client can start anomalyzer-mcp in the installed Python environment."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","criteria":[{"criterion":"Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","status":"covered","evidence":["src/anomalyzer/mcp.py:63","tests/test_mcp.py:27"],"notes":"Discovery exposes three read-only tools with the same strict argument contracts used at dispatch and advertised output schemas."},{"criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"covered","evidence":["src/anomalyzer/mcp.py:228","tests/test_mcp.py:248","tests/test_mcp.py:268"],"notes":"Legacy initialization and modern stateless discovery tests establish protocol metadata and evidence interpretation guidance."},{"criterion":"Analysis tool calls return structured content together with compact text.","status":"covered","evidence":["src/anomalyzer/mcp.py:162","tests/test_mcp.py:39"],"notes":"The manual conversion call returns a compact headline and structured run ID, counts, readiness, and preview."},{"criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"covered","evidence":["src/anomalyzer/mcp.py:156","src/anomalyzer/mcp.py:257","tests/test_mcp.py:235","tests/test_mcp.py:367"],"notes":"Tests reject invalid cursor/limit types, unknown fields, numeric run IDs, and missing session runs in both protocols without fabricated results."}],"ignore_findings":[],"done":true,"input_fingerprint":"7bd92d238f4f9545372c6c35e78aefea60c585bea0a4a8bb99096ab6cef27f96","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Connect a local server"},"impact":1,"gaps":0,"r":43,"x":903.3333333333334,"y":406},{"id":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","label":"Find values that break my ...","area":"investigation","scopes":["investigation","investigation/rules"],"scope":"investigation/rules","scopeLabel":"investigation/rules","status":"implemented","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","acceptance":["Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","Absolute and relative changes must exceed their declared maximum to violate a rule.","Relative-change checks require a nonzero previous value and do not bridge missing observations.","Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them."],"guidance":["An exact rule crossing does not establish cause, an incident, or business impact."],"path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","dependencies":["investigation/assessments/describe-early-changes"],"suggestedDependencies":[],"dependents":["investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"rule-boundaries","description":"Check inclusive range boundaries and strict absolute/relative change thresholds.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_explicit_rules_respect_boundaries"},{"id":"relative-availability","description":"Avoid relative violations across zero or missing values.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"explicit-not-actionable","description":"Create evidence and a case for an exact rule crossing without enabling action by default.","command":"python -m pytest -q tests/test_agentic.py::test_explicit_rule_is_exact_but_not_actionable_without_policy"},{"id":"readable-rule","description":"Identify the exact unit-cost range rule in CLI output.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the unit-cost example's crossed rule and confirm its text distinguishes the declared bound from statistical detection."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","criteria":[{"criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"covered","evidence":["src/anomalyzer/early.py:22","src/anomalyzer/contracts.py:209","tests/test_capability_behaviors.py:74"],"notes":"Boundary values 10 and 20 are accepted; 9 and 21 violate the declared range."},{"criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"covered","evidence":["src/anomalyzer/early.py:25","tests/test_capability_behaviors.py:74"],"notes":"Parameterized tests distinguish equality to the threshold from strict exceedance for both change rules."},{"criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"covered","evidence":["src/anomalyzer/early.py:29","src/anomalyzer/early.py:53","tests/test_capability_behaviors.py:87"],"notes":"Zero/missing predecessors suppress relative comparisons, and missing values reset comparison history."},{"criterion":"Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them.","status":"covered","evidence":["src/anomalyzer/early.py:34","src/anomalyzer/early.py:121","tests/test_agentic.py:49","tests/test_cli.py:134"],"notes":"Violation records retain rule ID/kind/bounds; default eligibility is false and replay requires explicit permission."}],"ignore_findings":[],"done":true,"input_fingerprint":"60d24e84faee9c1e2ae8f9337d17334b37c89305093404da2e84c51526111d0b","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Apply my stated limits"},"impact":1,"gaps":0,"r":43,"x":1086.6666666666667,"y":406},{"id":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","label":"Find changes in conversion...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","acceptance":["Ratios retain the computed value, formula, and original source indexes for each sample.","Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","Derived series can be analyzed with the frozen or adaptive single-resolution recipe."],"guidance":["Relationships are caller-supplied declarations, not discovered or causal relationships.","A difference requires identical declared units; a normalized residual requires a positive supplied scale.","A joint condition is a declared 0/1 comparison, not statistical proof."],"path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","dependencies":["relationships/quality/require-compatible-sources","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_examples.py"],"automatedChecks":[{"id":"ratio-lineage","description":"Explain a changing rate with the exact source indexes and formula.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"other-comparisons","description":"Check declared differences, normalized residuals, and combined conditions.","command":"python -m pytest -q tests/test_agentic.py::test_remaining_relationship_primitives"},{"id":"unavailable-values","description":"Keep zero-denominator and missing-source observations unavailable.","command":"python -m pytest -q tests/test_agentic.py::test_missing_and_zero_denominator_are_unavailable_not_zero"},{"id":"adaptive-relationship","description":"Use the adaptive recipe for a declared relationship.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_recipe_runs_for_declared_relationships"}],"manualChecks":["Inspect examples/agentic/conversion.json and distinguish its relationship departure from the individual source metrics."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","criteria":[{"criterion":"Ratios retain the computed value, formula, and original source indexes for each sample.","status":"covered","evidence":["src/anomalyzer/relationships.py:84","src/anomalyzer/relationships.py:152","tests/test_agentic.py:211"],"notes":"Assertions check exact ratio values, formula, and both original source indexes."},{"criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"covered","evidence":["src/anomalyzer/relationships.py:97","src/anomalyzer/relationships.py:134","tests/test_agentic.py:285"],"notes":"Parameterized assertions independently specify difference, normalized residual, all-condition, and any-condition outputs."},{"criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"covered","evidence":["src/anomalyzer/relationships.py:89","src/anomalyzer/relationships.py:142","tests/test_agentic.py:226"],"notes":"Lineage retains null values with missing_source_value/zero_denominator reasons. Explicit zero_denominator=error intentionally fails instead, as documented."},{"criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:87","src/anomalyzer/orchestrator.py:135","tests/test_agentic.py:171","tests/test_examples.py:17"],"notes":"Both supported recipes dispatch derived datasets through the shared single-resolution engine."}],"ignore_findings":[],"done":true,"input_fingerprint":"7fa8f7ffdc0d852ad961b3e1753761d1bd3b00b87b8c001a957ec19e26d068b9","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Compare ratios and metric gaps"},"impact":1,"gaps":0,"r":43,"x":1270,"y":406},{"id":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","label":"Find unusual spikes and dr...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","acceptance":["A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","Separated point departures remain separate episodes, and positional episode intervals use integer coordinates."],"guidance":["Detection is a numerical departure from the selected expectation, not a claim about cause or business impact."],"path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/evidence.py","tests/test_analysis.py"],"automatedChecks":[{"id":"point-evidence","description":"Detect an injected spike and retain deterministic earlier numerical evidence.","command":"python -m pytest -q tests/test_analysis.py::test_spike_and_no_lookahead"},{"id":"adjacent-run","description":"Describe adjacent point triggers and their contributing positions.","command":"python -m pytest -q tests/test_analysis.py::test_anomaly_stream_detects_consecutive_run"},{"id":"isolated-events","description":"Keep separated anomalies from becoming a consecutive-run pattern.","command":"python -m pytest -q tests/test_analysis.py::test_isolated_anomalies_do_not_form_second_order_pattern"},{"id":"positional-episodes","description":"Use integer intervals for position-only episode evidence.","command":"python -m pytest -q tests/test_analysis.py::test_positional_episode_uses_integer_interval"}],"manualChecks":["Inspect observations and triggering_samples for examples/spike.json and distinguish episode grouping from pattern findings."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","criteria":[{"criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"covered","evidence":["src/anomalyzer/recipes.py:344","src/anomalyzer/recipes.py:388","tests/test_analysis.py:441"],"notes":"The injected spike retains its observed value, expectation, standardized residual, and point trigger."},{"criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"covered","evidence":["src/anomalyzer/baseline.py:31","src/anomalyzer/recipes.py:174","tests/test_analysis.py:441","tests/test_trend.py:36"],"notes":"Assertions establish repeatability and unchanged earlier evidence under later evaluation changes with fixed seasonal settings. Automatic positional season discovery has the separately documented streaming limitation."},{"criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"covered","evidence":["src/anomalyzer/evidence.py:95","tests/test_analysis.py:274"],"notes":"The test checks one three-sample run, exact contributing positions, direction, and first detection."},{"criterion":"Separated point departures remain separate episodes, and positional episode intervals use integer coordinates.","status":"covered","evidence":["src/anomalyzer/evidence.py:12","src/anomalyzer/evidence.py:23","tests/test_analysis.py:263","tests/test_analysis.py:369"],"notes":"Separated spikes produce two episodes without a consecutive-run pattern; positional evidence produces interval [9, 9]."}],"ignore_findings":[],"done":true,"input_fingerprint":"062833cc1bd19d5734b0f3c63d1e81666ffb568ead3399e359d62db2c0c0677d","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Find spikes and drops"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","label":"Know when the expected bas...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","acceptance":["Repeating seasonal structure with stable calibration can yield supported readiness.","Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","Insufficient calibration support yields not_assessed, including when the run otherwise completes.","Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment."],"guidance":["Readiness does not prove real-world effectiveness and is not an automatic action-policy gate."],"path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/readiness.py","src/anomalyzer/recipes.py","tests/test_readiness.py"],"automatedChecks":[{"id":"supported-reference","description":"Support repeated seasonality and preserve readiness under later changes.","command":"python -m pytest -q tests/test_readiness.py::test_repeating_season_and_trend_support_reference"},{"id":"weak-season","description":"Warn on unstable seasonality even when no findings occur.","command":"python -m pytest -q tests/test_readiness.py::test_unstable_declared_season_warns_without_calling_series_inapplicable tests/test_readiness.py::test_documented_random_example_has_no_findings_but_warns_on_reference"},{"id":"calibration-drift","description":"Warn when residual calibration drifts despite a repeated training season.","command":"python -m pytest -q tests/test_readiness.py::test_calibration_drift_warns_even_when_training_season_repeats"},{"id":"not-assessed","description":"Distinguish uncalibrated history and too few calibration samples.","command":"python -m pytest -q tests/test_readiness.py::test_readiness_is_not_assessed_without_calibrated_evaluation tests/test_readiness.py::test_short_calibration_is_not_assessed_even_when_analysis_completes"}],"manualChecks":["Read readiness reasons alongside findings; a lack of findings alone does not establish a trustworthy reference."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","criteria":[{"criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"covered","evidence":["src/anomalyzer/readiness.py:128","tests/test_readiness.py:8"],"notes":"Assertions establish repeated seasonality, strong calibration improvement, and supported readiness."},{"criterion":"Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","status":"covered","evidence":["src/anomalyzer/readiness.py:113","src/anomalyzer/readiness.py:122","tests/test_readiness.py:28","tests/test_readiness.py:41"],"notes":"Completed analyses retain caution reasons for weak seasonal repetition or calibration-center drift."},{"criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"covered","evidence":["src/anomalyzer/readiness.py:77","tests/test_readiness.py:56","tests/test_readiness.py:63"],"notes":"Both uncalibrated history and a completed run with three calibration samples return not_assessed."},{"criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"covered","evidence":["src/anomalyzer/recipes.py:327","tests/test_readiness.py:8"],"notes":"Readiness reads training/calibration inputs once; tests compare identical segment and aggregate assessments after a later shift."}],"ignore_findings":[],"done":true,"input_fingerprint":"d01367627e058ea39c78ee0e3616061a2e0725d06e8acd0b84b46a9655431e57","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Assess baseline quality"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","label":"Follow evolving growth whi...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","acceptance":["Each rolling expectation uses only observations before the sample being scored.","Persistent growth changes can be followed more closely than by the frozen default recipe.","An isolated extreme value is held out of future model references rather than creating a seasonal echo.","Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected."],"guidance":["A persistent third extreme miss in one direction may be admitted to the rolling reference; this is not a manual regime reset."],"path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"causal-window","description":"Keep a known rolling expectation invariant when future values are appended.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_expectation_is_one_step_and_prefix_invariant"},{"id":"changing-growth","description":"Follow a persistent growth change and report model admissions.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_tracks_a_persistent_growth_change"},{"id":"isolated-extreme","description":"Exclude an isolated extreme from later reference estimates.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_contains_an_isolated_extreme"},{"id":"bounded-settings","description":"Reject conflicting or unbounded adaptive settings.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_settings_reject_conflicting_or_unbounded_trend"}],"manualChecks":["Inspect the adaptive recipe's rolling-window parameters and frozen calibration before interpreting its final expectation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","criteria":[{"criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"covered","evidence":["src/anomalyzer/baseline.py:61","tests/test_analysis.py:48"],"notes":"All rolling-reference positions precede the scored index. The independent expectation remains unchanged after future values are appended."},{"criterion":"Persistent growth changes can be followed more closely than by the frozen default recipe.","status":"covered","evidence":["src/anomalyzer/recipes.py:145","src/anomalyzer/recipes.py:370","tests/test_analysis.py:58"],"notes":"The test requires adaptive held-out MAE below half the frozen recipe's MAE and a close final expectation."},{"criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","tests/test_analysis.py:81"],"notes":"The isolated spike is excluded from model references; its next seasonal expectation closely matches the control. Persistent third misses are admitted as guidance explicitly permits."},{"criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:41","tests/test_analysis.py:104"],"notes":"Both incompatible explicit trend selection and excessive slope lookback raise specific validation errors."}],"ignore_findings":[],"done":true,"input_fingerprint":"984422f1fd3ab3d22cf7b4ba9d0bd06629fc228a8e53d7ad44439cd8d35ec0db","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Follow evolving growth"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","label":"Account for daily and week...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"planned","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","acceptance":["Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","Insufficient support for a requested cycle produces an explicit limitation or declared fallback rather than a silently fitted component.","Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle."],"guidance":["Begin with a bounded daily-and-weekly use case; overlapping cycles share information and cannot be treated as independent evidence.","Proposed automated checks cover combined ordinary cycles, future isolation, unsupported cycle history, and extreme-value influence across cycles.","Manual review should inspect a weekday morning spike and verify the baseline explains ordinary calendar variation without explaining away the spike."],"path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Represent more than one repeating cycle"},"impact":0,"gaps":3,"r":47,"x":170,"y":406},{"id":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","label":"Limit how extreme values d...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","acceptance":["Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","Training and calibration diagnostics identify excluded extreme observations.","An extreme evaluated sample remains evidence while its expected value is used in future references.","Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior."],"guidance":["Protection keeps the observed evidence; it does not delete observations or certify that a departure was an incident."],"path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/robust.py","src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py","tests/test_cli.py"],"automatedChecks":[{"id":"robust-reference","description":"Limit an extreme weekly reference and use earlier references at lag one.","command":"python -m pytest -q tests/test_analysis.py::test_robust_weekly_reference_limits_one_high_week tests/test_analysis.py::test_robust_lag_one_uses_prior_seasonal_values"},{"id":"training-calibration","description":"Record training and calibration exclusions and retain a useful growth fit.","command":"python -m pytest -q tests/test_trend.py::test_robust_training_fit_and_calibration_exclusions"},{"id":"include-choice","description":"Compare robust and include modes through the CLI and show their differing triggers.","command":"python -m pytest -q tests/test_cli.py::test_outlier_handling_cli"}],"manualChecks":["Inspect excluded_from_model, reference_action, and outlier-handling diagnostics alongside the original observed value."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","criteria":[{"criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"covered","evidence":["src/anomalyzer/baseline.py:39","tests/test_analysis.py:25"],"notes":"Assertions check a independently computed weighted reference and invariance when an already-trimmed high value becomes more extreme."},{"criterion":"Training and calibration diagnostics identify excluded extreme observations.","status":"covered","evidence":["src/anomalyzer/trend.py:178","src/anomalyzer/recipes.py:287","tests/test_trend.py:65"],"notes":"Tests identify exact training/calibration exclusions and compare their influence on trend selection and residual scale."},{"criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","src/anomalyzer/recipes.py:392","tests/test_analysis.py:263","tests/test_trend.py:11"],"notes":"The observed departure and trigger survive; only reference_values receives the prior expectation, preventing the next seasonal echo."},{"criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"covered","evidence":["src/anomalyzer/recipes.py:151","src/anomalyzer/recipes.py:475","tests/test_cli.py:95"],"notes":"CLI assertions distinguish robust triggers {65} from include-mode triggers {65, 72}."}],"ignore_findings":[],"done":true,"input_fingerprint":"183b8696fdc9786878b3e54c4a9e5f68180b47e157528bef953274a3487d49a7","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Protect later comparisons"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","label":"Choose the repeating cycle...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","acceptance":["An explicit season length is honored and reported as explicit.","A repeating positional series can infer a seasonal lag and report the selection window.","A positional series without a sufficiently stable candidate falls back to lag one and records why.","The reserved calibration and evaluation tail does not determine positional season selection."],"guidance":["Candidate discovery uses the input prefix before the reserved tail and confirms the candidate in an early window. It is not fully prefix-invariant streaming discovery.","Supply season_length explicitly when stable streaming season selection is required."],"path":".capabilities/analysis/expectations/select-season-length.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/seasonality.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"declared-season","description":"Honor a separate explicit season setting.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"inference-fallback","description":"Infer a repeating lag or record lag-one fallback when no season is supported.","command":"python -m pytest -q tests/test_analysis.py::test_positional_season_inference_and_fallback"},{"id":"reserved-tail","description":"Keep selection unchanged when only the reserved tail changes.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_season_inference_excludes_reserved_tail"},{"id":"cli-inference","description":"Expose the inferred lag and its source for bare JSON values in the CLI.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"}],"manualChecks":["Review selection_window and ranked candidates for a positional series before treating its inferred season as an ongoing monitoring assumption."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","criteria":[{"criterion":"An explicit season length is honored and reported as explicit.","status":"covered","evidence":["src/anomalyzer/recipes.py:55","tests/test_analysis.py:234"],"notes":"Explicit field presence bypasses inference, and the test checks the reported source."},{"criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"covered","evidence":["src/anomalyzer/seasonality.py:96","tests/test_analysis.py:247"],"notes":"The repeating seven-position series selects lag 7 and reports confirmation window [0, 27]."},{"criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"covered","evidence":["src/anomalyzer/seasonality.py:79","src/anomalyzer/seasonality.py:84","tests/test_analysis.py:247"],"notes":"Candidate-strength and early-confirmation failures return lag 1 with a reason; the quadratic example verifies fallback."},{"criterion":"The reserved calibration and evaluation tail does not determine positional season selection.","status":"covered","evidence":["src/anomalyzer/seasonality.py:69","tests/test_capability_behaviors.py:147"],"notes":"Discovery excludes calibration_size plus one reserved sample. Changing that tail preserves the complete inference diagnostics; arbitrary appended-prefix invariance is explicitly outside the guidance."}],"ignore_findings":[],"done":true,"input_fingerprint":"3317b1727dc3cabfcfc9b87808d9249d0d59f111b8fd5f8a1fad653b21343980","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Choose a repeating cycle"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","label":"Interpret rate changes wit...","area":"analysis","scopes":["analysis","analysis/rates"],"scope":"analysis/rates","scopeLabel":"analysis/rates","status":"planned","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","acceptance":["Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","Zero denominators remain unavailable, and low support is explicitly marked; intervals remain within the valid zero-to-one rate domain.","Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations."],"guidance":["Independence and stable exposure are modeling assumptions; clustered events, overdispersion, or changing populations require explicit limitations.","Proposed automated checks cover zero denominators, tiny counts, zero-or-all-success boundaries, and equal rates at different volumes.","Manual review should compare sparse and high-volume conversion changes and confirm the explanation discloses assumptions without inventing incident probability."],"path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Judge rates using their exposure"},"impact":0,"gaps":3,"r":47,"x":194,"y":524},{"id":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","label":"See changes before the day...","area":"analysis","scopes":["analysis","analysis/resolution"],"scope":"analysis/resolution","scopeLabel":"analysis/resolution","status":"implemented","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","acceptance":["Callers can select sum, mean, or last aggregation for completed and partial periods.","Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","An explicitly unfinished source sample is excluded from both analysis views.","A source cadence that does not evenly divide 24 hours is rejected."],"guidance":["Periods are fixed 24-hour windows, not daylight-saving-aware local calendar days.","Native and daily views share source observations and are correlated.","Legacy schema 1.0 requires complete input and does not support resets. Schema 1.1 labels sparse native/completed-day views and preserves reset boundaries; daily sums require declared additive volume and fully observed children."],"path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","dependencies":["data/quality/exclude-incomplete-periods","analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"selected-aggregation","description":"Check sum, mean, and last for completed days and trailing partial aggregates.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_multi_resolution_uses_selected_aggregation"},{"id":"two-views","description":"Return completed days, native evidence, and a linked trailing partial-day marker.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_current_period_and_keeps_two_views"},{"id":"unfinished-source","description":"Exclude a marked incomplete source sample from both views.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days"},{"id":"dividing-cadence","description":"Reject subperiod durations that do not evenly divide a day.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_rejects_non_dividing_frequency"}],"manualChecks":["Inspect the partial aggregate's observed_aggregate and intraday_evidence_refs together with the two method namespaces."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","criteria":[{"criterion":"Callers can select sum, mean, or last aggregation for completed and partial periods.","status":"covered","evidence":["src/anomalyzer/contracts.py:18","src/anomalyzer/recipes.py:539","tests/test_capability_behaviors.py:178"],"notes":"Parameterized assertions check independently known completed and partial values for all three aggregation choices."},{"criterion":"Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","status":"covered","evidence":["src/anomalyzer/recipes.py:617","src/anomalyzer/recipes.py:667","src/anomalyzer/recipes.py:680","tests/test_analysis.py:170"],"notes":"Separate datasets, configurations, and namespaces preserve the two correlated views."},{"criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"covered","evidence":["src/anomalyzer/recipes.py:639","src/anomalyzer/recipes.py:690","tests/test_analysis.py:170"],"notes":"The marker contains finalized progress, observed aggregate, native evidence references, and an explicit exclusion flag."},{"criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"covered","evidence":["src/anomalyzer/validation.py:44","src/anomalyzer/recipes.py:564","tests/test_analysis.py:200"],"notes":"Both child datasets use prepared finalized observations; tests verify counts and retention of the excluded original value."},{"criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"covered","evidence":["src/anomalyzer/recipes.py:572","tests/test_analysis.py:221"],"notes":"A five-hour cadence raises the specific dividing-cadence error."}],"ignore_findings":[],"done":true,"input_fingerprint":"cdcdea1166a4e3d2523b594bec938142135e17a6b4279732e52819a486d1b4f2","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Compare finalized subperiods"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","label":"Compare days and weeks in ...","area":"data","scopes":["data","data/calendar"],"scope":"data/calendar","scopeLabel":"data/calendar","status":"planned","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","acceptance":["Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","Periods report UTC boundaries and local labels; daylight-saving days can contain 23 or 25 hours without being forced into fixed 24-hour windows.","Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation."],"guidance":["Calendar grouping needs a distinct contract from fixed elapsed cadence; it must not silently reinterpret existing frequency or aggregation settings.","Proposed automated checks cover spring transitions, fall transitions, week boundaries, and missing native slots in a calendar period.","Manual review should inspect a daylight-saving week and confirm local labels, observed coverage, and completed-versus-open comparisons are understandable."],"path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Declare local calendar boundaries"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","label":"Analyze many entities with...","area":"data","scopes":["data","data/cohorts"],"scope":"data/cohorts","scopeLabel":"data/cohorts","status":"planned","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","acceptance":["Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","Training, calibration, relationships, and cases do not cross entity boundaries unless a separate explicit aggregate comparison is requested.","The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident."],"guidance":["Batch orchestration should preserve existing per-request contracts rather than silently lifting schema limits; multiple-testing limits stay visible.","Proposed automated checks cover partition isolation, batch-versus-single equivalence, one-entity failure, and shared resource-budget exhaustion.","Manual review should inspect a mixed batch of healthy, changed, and sparse entities and confirm the summary does not equate missing support with health."],"path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Keep each entity's evidence separate"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","label":"Inspect distinct events sh...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Callers supply stable logical event IDs so distinct comparable calls at one timestamp retain separate values and evidence without inventing chronology.","intent":"Make real per-call evidence consumable while protecting references, replay and pattern interpretation from retry duplication and arbitrary tie order.","acceptance":["Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results."],"guidance":["Callers own identity authenticity, retry deduplication, compatible populations and any assertion of trustworthy sequence. Sorting an ID establishes no cause.","Cases and overlapping findings are inspectable numerical evidence; applications own incident deduplication, business severity and notifications."],"path":".capabilities/data/quality/identify-distinct-events.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","data/quality/analyze-sparse-observations","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/events.py","src/anomalyzer/sparse.py","src/anomalyzer/orchestrator.py","src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/cases.py","src/anomalyzer/mcp_views.py","tests/test_events.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"identity-and-retries","description":"Retain distinct tied calls and reject duplicate logical identities.","command":"python -m pytest -q tests/test_events.py -k \"identity or repeat or distinct or reordering\""},{"id":"honest-order","description":"Keep point/range evidence and require actual declared order for tied sequences.","command":"python -m pytest -q tests/test_events.py -k \"cohort or ties or trusted or reset or lag\""},{"id":"reproducible-replay","description":"Preserve prefix, as-of, addressed corrections and case revisions.","command":"python -m pytest -q tests/test_events.py -k \"prefix or asof or correction or revision or reference\""},{"id":"public-consumers","description":"Validate schemas, Python, CLI, MCP retrieval and bounded response growth.","command":"python -m pytest -q tests/test_events.py -k \"python_schema or bounded\""}],"manualChecks":["Inspect progression figures 23–24 and the API guide for exact original times, layout versus trusted order, retry responsibility and overlapping evidence limits."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/identify-distinct-events.","criteria":[{"criterion":"Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","status":"partial","evidence":["src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:118","src/anomalyzer/contracts.py:192","src/anomalyzer/contracts.py:193","src/anomalyzer/contracts.py:241","src/anomalyzer/events.py:1","src/anomalyzer/events.py:14","src/anomalyzer/events.py:22","src/anomalyzer/events.py:23","src/anomalyzer/events.py:26","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:46","src/anomalyzer/sparse.py:60","src/anomalyzer/sparse.py:62","src/anomalyzer/orchestrator.py:21","src/anomalyzer/orchestrator.py:84","src/anomalyzer/orchestrator.py:90","src/anomalyzer/orchestrator.py:94","src/anomalyzer/orchestrator.py:100","src/anomalyzer/alignment.py:14","src/anomalyzer/alignment.py:33","src/anomalyzer/alignment.py:46","src/anomalyzer/alignment.py:47","src/anomalyzer/alignment.py:48","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:35","src/anomalyzer/cases.py:16","src/anomalyzer/cases.py:18","src/anomalyzer/cases.py:20","src/anomalyzer/cases.py:22","src/anomalyzer/cases.py:24","src/anomalyzer/mcp_views.py:68","src/anomalyzer/mcp_views.py:71","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:77","src/anomalyzer/mcp_views.py:89","tests/test_events.py:1","tests/test_events.py:22","tests/test_events.py:24","tests/test_events.py:25","tests/test_events.py:41","docs/sparse-data.md:50","docs/sparse-data.md:73","docs/sparse-data.md:74","docs/sparse-data.md:77","docs/sparse-data.md:83","examples/sparse_examples.py:36","examples/sparse_examples.py:57","examples/sparse_examples.py:65","examples/sparse_examples.py:66","examples/sparse_examples.py:69","examples/render_readme_progression.py:141","examples/render_readme_progression.py:301","examples/render_readme_progression.py:303","examples/render_readme_progression.py:373","examples/render_readme_progression.py:413"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:145","src/anomalyzer/contracts.py:246","src/anomalyzer/contracts.py:255","src/anomalyzer/contracts.py:256","src/anomalyzer/events.py:26","src/anomalyzer/events.py:27","src/anomalyzer/events.py:28","src/anomalyzer/events.py:37","src/anomalyzer/events.py:46","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse.py:60","src/anomalyzer/orchestrator.py:11","src/anomalyzer/orchestrator.py:39","src/anomalyzer/orchestrator.py:130","src/anomalyzer/orchestrator.py:144","src/anomalyzer/orchestrator.py:170","src/anomalyzer/alignment.py:62","src/anomalyzer/relationships.py:38","src/anomalyzer/relationships.py:44","src/anomalyzer/relationships.py:164","src/anomalyzer/relationships.py:166","src/anomalyzer/relationships.py:168","src/anomalyzer/cases.py:11","src/anomalyzer/cases.py:25","src/anomalyzer/cases.py:31","src/anomalyzer/cases.py:44","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:188","src/anomalyzer/mcp_views.py:304","src/anomalyzer/mcp_views.py:322","src/anomalyzer/mcp_views.py:326","tests/test_events.py:18","tests/test_events.py:23","tests/test_events.py:30","tests/test_events.py:47","tests/test_events.py:56","docs/sparse-data.md:31","docs/sparse-data.md:88","docs/sparse-data.md:93","docs/sparse-data.md:106","docs/sparse-data.md:111","examples/sparse_examples.py:65","examples/sparse_examples.py:67","examples/sparse_examples.py:74","examples/sparse_examples.py:81","examples/sparse_examples.py:83","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349","examples/render_readme_progression.py:352","examples/render_readme_progression.py:477","examples/render_readme_progression.py:542"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","status":"partial","evidence":["src/anomalyzer/contracts.py:246","src/anomalyzer/contracts.py:255","src/anomalyzer/contracts.py:256","src/anomalyzer/contracts.py:268","src/anomalyzer/contracts.py:296","src/anomalyzer/events.py:1","src/anomalyzer/events.py:50","src/anomalyzer/events.py:71","src/anomalyzer/events.py:80","src/anomalyzer/events.py:84","src/anomalyzer/sparse.py:46","src/anomalyzer/sparse.py:62","src/anomalyzer/sparse.py:93","src/anomalyzer/sparse.py:170","src/anomalyzer/sparse.py:182","src/anomalyzer/orchestrator.py:24","src/anomalyzer/orchestrator.py:142","src/anomalyzer/orchestrator.py:193","src/anomalyzer/orchestrator.py:209","src/anomalyzer/orchestrator.py:213","src/anomalyzer/alignment.py:64","src/anomalyzer/alignment.py:66","src/anomalyzer/alignment.py:68","src/anomalyzer/relationships.py:38","src/anomalyzer/relationships.py:198","src/anomalyzer/relationships.py:205","src/anomalyzer/cases.py:59","src/anomalyzer/cases.py:62","src/anomalyzer/cases.py:115","src/anomalyzer/cases.py:118","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:152","src/anomalyzer/mcp_views.py:380","src/anomalyzer/mcp_views.py:385","tests/test_events.py:29","tests/test_events.py:30","tests/test_events.py:80","tests/test_events.py:81","tests/test_events.py:121","docs/sparse-data.md:118","docs/sparse-data.md:136","docs/sparse-data.md:137","docs/sparse-data.md:138","docs/sparse-data.md:139","examples/sparse_examples.py:87","examples/sparse_examples.py:88","examples/sparse_examples.py:89","examples/render_readme_progression.py:918","examples/render_readme_progression.py:921","examples/render_readme_progression.py:1024","examples/render_readme_progression.py:1037","examples/render_readme_progression.py:1038"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results.","status":"partial","evidence":["src/anomalyzer/contracts.py:308","src/anomalyzer/contracts.py:315","src/anomalyzer/contracts.py:319","src/anomalyzer/contracts.py:596","src/anomalyzer/contracts.py:597","src/anomalyzer/events.py:1","src/anomalyzer/events.py:14","src/anomalyzer/events.py:100","src/anomalyzer/events.py:102","src/anomalyzer/events.py:165","src/anomalyzer/sparse.py:182","src/anomalyzer/sparse.py:194","src/anomalyzer/sparse.py:195","src/anomalyzer/sparse.py:202","src/anomalyzer/sparse.py:333","src/anomalyzer/orchestrator.py:122","src/anomalyzer/orchestrator.py:143","src/anomalyzer/orchestrator.py:155","src/anomalyzer/orchestrator.py:162","src/anomalyzer/orchestrator.py:164","src/anomalyzer/alignment.py:63","src/anomalyzer/alignment.py:66","src/anomalyzer/relationships.py:35","src/anomalyzer/relationships.py:198","src/anomalyzer/relationships.py:203","src/anomalyzer/relationships.py:205","src/anomalyzer/cases.py:29","src/anomalyzer/cases.py:30","src/anomalyzer/cases.py:33","src/anomalyzer/cases.py:37","src/anomalyzer/cases.py:62","src/anomalyzer/mcp_views.py:12","src/anomalyzer/mcp_views.py:14","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:77","src/anomalyzer/mcp_views.py:88","tests/test_events.py:1","tests/test_events.py:12","tests/test_events.py:36","tests/test_events.py:49","tests/test_events.py:51","docs/sparse-data.md:37","docs/sparse-data.md:97","docs/sparse-data.md:100","docs/sparse-data.md:111","docs/sparse-data.md:115","examples/sparse_examples.py:39","examples/sparse_examples.py:72","examples/sparse_examples.py:73","examples/sparse_examples.py:77","examples/sparse_examples.py:80","examples/render_readme_progression.py:3","examples/render_readme_progression.py:781","examples/render_readme_progression.py:782","examples/render_readme_progression.py:787","examples/render_readme_progression.py:798"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","partial: Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","partial: Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","partial: Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","impact":0,"gaps":0,"r":38,"x":377.33333333333337,"y":524},{"id":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","label":"Know when missing updates ...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"planned","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","acceptance":["Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","Results distinguish current, late, missing, and not-assessed freshness and expose the last observed update and the deadline used.","A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision."],"guidance":["Freshness depends on declared collection expectations; it does not prove source completeness and does not start a scheduler or send a notification.","Proposed automated checks cover exact deadlines, allowed lateness, absent source updates, and stale-source propagation to relationships.","Manual review should compare a stale feed with a true metric drop and confirm the report distinguishes collection uncertainty from numerical departure."],"path":".capabilities/data/quality/report-source-freshness.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Check whether the sources are current"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","label":"See how late or corrected ...","area":"data","scopes":["data","data/revisions"],"scope":"data/revisions","scopeLabel":"data/revisions","status":"planned","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","acceptance":["A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","Results preserve prior and new evidence versions and identify findings that were added, changed, or withdrawn with traceable source lineage.","Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results."],"guidance":["Schema-1.1 snapshot revision selection and as-of availability are implemented in data/quality/analyze-sparse-observations; durable prior/new evidence diffing and automatic reconciliation in this capability remain planned.","A corrected event-time result does not replace the record of what was known earlier; callers control retention and any notification about a revision.","Proposed automated checks cover late completion, corrected calibration data, downstream relationship changes, and conflicting source versions.","Manual review should inspect a drop withdrawn after ingestion completes and confirm both the original limitation and the revision explanation are retained."],"path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Reconcile revisions without losing history"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","label":"Choose settings using held...","area":"evaluation","scopes":["evaluation","evaluation/comparisons"],"scope":"evaluation/comparisons","scopeLabel":"evaluation/comparisons","status":"planned","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","acceptance":["Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","The output preserves every candidate's settings and evaluation provenance; adopting a configuration requires an explicit caller decision."],"guidance":["Bounded comparison is not unrestricted parameter search; held-out results describe the supplied history and do not guarantee future effectiveness.","Proposed automated checks cover split isolation, comparable replay inputs, failed candidate retention, and unchanged active settings after comparison.","Manual review should compare an earlier but noisier configuration with a quieter one and confirm the tradeoff is visible without an unsupported winner."],"path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Review the tradeoff before adopting settings"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","label":"Measure detection timing a...","area":"evaluation","scopes":["evaluation","evaluation/outcomes"],"scope":"evaluation/outcomes","scopeLabel":"evaluation/outcomes","status":"planned","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","acceptance":["Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","Timing reports distinguish the incident's supplied start from the first available candidate, calibrated detection, and policy-eligible evidence.","Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown."],"guidance":["Detection timing does not measure response effectiveness or establish that an unmatched finding was false; incident labels are caller judgments.","Proposed automated checks cover boundary matching, ambiguous overlaps, incomplete label coverage, and separate candidate versus eligibility timing.","Manual review should inspect a matched and an unmatched episode and confirm the reported timing and accuracy limits agree with the supplied label coverage."],"path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Compare detection with reviewed outcomes"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","label":"Replay history as it would...","area":"evaluation","scopes":["evaluation","evaluation/replay"],"scope":"evaluation/replay","scopeLabel":"evaluation/replay","status":"planned","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","acceptance":["Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","Repeating a replay preserves numerical evidence and first-detection timing apart from generated IDs and runtime; stopped replays retain completed steps."],"guidance":["Event-time replay does not reconstruct arrival-time availability unless the caller supplies versioned arrival history.","Proposed automated checks cover future-tail isolation, short-history cutoffs, deterministic repeated replay, and interrupted replay retention.","Manual review should inspect one change from its first available observation through calibration and confirm the explanation uses only then-known evidence."],"path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Reconstruct the evidence available at the time"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","label":"Update evidence when the n...","area":"execution","scopes":["execution","execution/incremental"],"scope":"execution/incremental","scopeLabel":"execution/incremental","status":"planned","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","acceptance":["Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","Supported explicitly configured recipes produce the same numerical evidence as full batch analysis of the combined history, including cross-chunk patterns.","Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations."],"guidance":["Start with explicit season lengths; input-prefix season discovery is not currently guaranteed to be invariant when the history grows.","Proposed automated checks cover chunk-versus-batch equivalence, cross-chunk CUSUM, idempotent retry, and failed-append state preservation.","Manual review should follow a sustained change across several append batches and confirm detection timing and provisional evidence remain understandable."],"path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Analyze the next finalized samples"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","label":"Keep useful evidence when ...","area":"execution","scopes":["execution","execution/resilience"],"scope":"execution/resilience","scopeLabel":"execution/resilience","status":"implemented","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","acceptance":["A method failure preserves evidence completed before the failure and reports the error.","Runtime budgets are checked cooperatively and produce an explicit stop reason.","CLI byte limits and execution failures produce distinct error exits.","A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial."],"guidance":["Default limits are 10,000 source observations per request, 10 MB CLI input, and 60 seconds cooperative runtime.","Numerical results are deterministic for fixed declared settings; generated run IDs and runtimes vary."],"path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/budget.py","src/anomalyzer/recipes.py","src/anomalyzer/orchestrator.py","src/anomalyzer/io.py","src/anomalyzer/cli.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"method-and-budget","description":"Retain pre-failure evidence and report a runtime-budget stop.","command":"python -m pytest -q tests/test_analysis.py::test_failure_and_budget"},{"id":"cli-limits","description":"Enforce byte limits and distinguish execution failure in CLI output and exit status.","command":"python -m pytest -q tests/test_cli.py::test_size_limit_and_execution_exit"},{"id":"partial-siblings","description":"Preserve healthy analysis when a sibling dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Inspect run status, stop_reason, and each target's status before relying on a result; confirm limits are presented as cooperative rather than hard process timeouts."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","criteria":[{"criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"covered","evidence":["src/anomalyzer/recipes.py:400","src/anomalyzer/recipes.py:448","tests/test_analysis.py:509"],"notes":"The failure regression preserves evidence through index 44 and reports the injected error at index 45."},{"criterion":"Runtime budgets are checked cooperatively and produce an explicit stop reason.","status":"covered","evidence":["src/anomalyzer/budget.py:18","src/anomalyzer/recipes.py:108","src/anomalyzer/recipes.py:508","tests/test_analysis.py:509"],"notes":"Cooperative target/sample checks expose runtime_budget and budget_exceeded; this is not a hard process timeout."},{"criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:429","tests/test_cli.py:108"],"notes":"The full suite executed the byte-limit and execution-exit regression successfully; invalid input exits 2 and execution failure exits 1."},{"criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:125","src/anomalyzer/orchestrator.py:171","src/anomalyzer/orchestrator.py:185","tests/test_capability_behaviors.py:159","tests/test_capability_behaviors.py:30"],"notes":"Healthy dataset evidence survives target-specific failure; inspected incomplete-usage reports partial with no invented healthy relationship."}],"ignore_findings":[],"done":true,"input_fingerprint":"146a9c8396fd7a2d039f75077dc0b82a0778685d20c3782aa284d5f0b6ed94be","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect incomplete execution"},"impact":0,"gaps":0,"r":38,"x":720,"y":642},{"id":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","label":"Resume local analysis afte...","area":"execution","scopes":["execution","execution/state"],"scope":"execution/state","scopeLabel":"execution/state","status":"planned","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","acceptance":["A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","Restoring a supported state and continuing yields the same numerical evidence and detection positions as uninterrupted processing of the same observations.","Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled."],"guidance":["Stored observations and context can contain sensitive data; document exactly what a snapshot contains and avoid automatic persistence.","Proposed automated checks cover restart equivalence, corrupt snapshots, version or settings mismatch, and output overwrite protection.","Manual review should inspect a snapshot and resumed explanation and confirm their provenance and retained-data scope are clear to an application owner."],"path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Preserve state across restarts"},"impact":0,"gaps":3,"r":47,"x":744,"y":760},{"id":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","label":"Understand findings in a r...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","acceptance":["Point-anomaly episode counts and residual-pattern findings are labeled separately.","Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","Readiness status remains visible, including when it was not assessed."],"guidance":["Readable summaries should retain the evidence's claim limits and allow drilldown into JSON."],"path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/cli.py","tests/test_cli.py","examples/agentic/conversion.json","examples/agentic/unit-cost.json"],"automatedChecks":[{"id":"episodes-patterns","description":"Distinguish shift patterns from point episodes in readable output.","command":"python -m pytest -q tests/test_cli.py::test_readable_output_distinguishes_episodes_from_patterns"},{"id":"relationship-case","description":"Explain a conversion case with entity, units, comparison, and a suggested check.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_relationship_request_through_cli"},{"id":"explicit-rule","description":"Describe the unit-cost rule boundary and unassessed readiness.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the conversion and unit-cost example output and confirm the descriptions state the supported numerical comparison in plain language."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","criteria":[{"criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"covered","evidence":["src/anomalyzer/cli.py:347","src/anomalyzer/cli.py:405","tests/test_cli.py:143"],"notes":"Readable tests distinguish zero-episode residual shifts from a spike episode and retain the overlap warning."},{"criterion":"Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","status":"covered","evidence":["src/anomalyzer/cli.py:291","src/anomalyzer/cli.py:308","src/anomalyzer/cli.py:331","tests/test_cli.py:117","examples/agentic/conversion.json"],"notes":"Conversion output identifies site=demo, observed/expected orders_per_visit, the relationship, and a collection-completeness investigation question."},{"criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"covered","evidence":["src/anomalyzer/cli.py:318","tests/test_cli.py:134","examples/agentic/unit-cost.json"],"notes":"Unit-cost output names review-above-13c and its [0, 0.13] boundary as a declared rule crossing."},{"criterion":"Readiness status remains visible, including when it was not assessed.","status":"covered","evidence":["src/anomalyzer/cli.py:352","src/anomalyzer/cli.py:399","tests/test_cli.py:134"],"notes":"Readable output includes not_assessed and its supporting reasons without claiming measured alert accuracy."}],"ignore_findings":[],"done":true,"input_fingerprint":"98108868b000972fd9b3c3f082846fbd3f3b08c399ef7ed424c23245d6935fef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Read the findings"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","label":"Share an investigation tha...","area":"interfaces","scopes":["interfaces","interfaces/export"],"scope":"interfaces/export","scopeLabel":"interfaces/export","status":"planned","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","acceptance":["Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","Source observations and context are included only by explicit selection; omitted data and resulting limits on independent reproduction are disclosed."],"guidance":["Fingerprints identify included provenance; they do not prove source truth. A portable report must make privacy scope and reproduction limits inspectable.","Proposed automated checks cover manifest consistency, partial-run export, omitted-source disclosure, and overwrite protection.","Manual review should open a bundle outside the original session and confirm a reviewer can trace one finding and see exactly what cannot be reproduced."],"path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Preserve a portable evidence record"},"impact":0,"gaps":3,"r":47,"x":927.3333333333334,"y":760},{"id":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","label":"Install a versioned analys...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","acceptance":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"guidance":["Guard owns scheduling, persistence, business severity, incidents and notifications.","A published package does not establish integration into UsageTap Guard."],"path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","interfaces/cli/analyze-files-and-streams","interfaces/mcp/connect-read-only-local-server"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["pyproject.toml","src/anomalyzer/_version.py","scripts/release.py","scripts/publish_release.py","scripts/verify_distribution.py","scripts/verify_lambda.sh","examples/library/guard_smoke.py",".github/workflows/ci.yml",".github/workflows/release.yml","docs/releases/README.md","tests/test_release.py"],"automatedChecks":[{"id":"release-safety","description":"Reject mismatched tags, altered assets and mutable publication.","command":"python -m pytest -q tests/test_release.py"},{"id":"installed-artifacts","description":"Install wheel and sdist outside the checkout and invoke public APIs.","command":"python scripts/verify_distribution.py --python 3.12"},{"id":"release-inventory","description":"Check archive contents, version metadata and SHA-256 inventory.","command":"python scripts/release.py verify"},{"id":"linux-offline-consumer","description":"Bundle and run with networking disabled in Linux Python 3.12.","command":"bash scripts/verify_lambda.sh"}],"manualChecks":["Follow the release guide and confirm Guard retains domain severity and incident ownership; inspect readiness and evidence limits before adopting."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for interfaces/python/install-versioned-library.","criteria":[{"criterion":"Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","status":"partial","evidence":["src/anomalyzer/_version.py:1","scripts/release.py:1","scripts/release.py:20","scripts/release.py:39","scripts/release.py:48","scripts/release.py:54","scripts/publish_release.py:21","scripts/publish_release.py:34","scripts/publish_release.py:42","scripts/publish_release.py:53","scripts/verify_lambda.sh:15","scripts/verify_lambda.sh:18","scripts/verify_lambda.sh:20","scripts/verify_lambda.sh:26","scripts/verify_lambda.sh:28","examples/library/guard_smoke.py:74",".github/workflows/ci.yml:14",".github/workflows/ci.yml:19",".github/workflows/ci.yml:27",".github/workflows/ci.yml:43",".github/workflows/ci.yml:48",".github/workflows/release.yml:31",".github/workflows/release.yml:52",".github/workflows/release.yml:61","docs/releases/README.md:16","docs/releases/README.md:17","docs/releases/README.md:18","docs/releases/README.md:21","docs/releases/README.md:22","tests/test_release.py:60","tests/test_release.py:71","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","status":"partial","evidence":["scripts/verify_distribution.py:1","scripts/verify_distribution.py:48","examples/library/guard_smoke.py:32","examples/library/guard_smoke.py:49","examples/library/guard_smoke.py:65","examples/library/guard_smoke.py:69","examples/library/guard_smoke.py:76",".github/workflows/ci.yml:59",".github/workflows/release.yml:60","docs/releases/README.md:17","docs/releases/README.md:99","docs/releases/README.md:217","docs/releases/README.md:220","docs/releases/README.md:222"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:34","scripts/release.py:81","scripts/release.py:82","scripts/release.py:85","scripts/release.py:92","scripts/publish_release.py:1","scripts/publish_release.py:18","scripts/publish_release.py:19","scripts/publish_release.py:20","scripts/publish_release.py:40","scripts/verify_distribution.py:47","scripts/verify_lambda.sh:25","examples/library/guard_smoke.py:39",".github/workflows/ci.yml:44",".github/workflows/ci.yml:47",".github/workflows/ci.yml:48",".github/workflows/ci.yml:51",".github/workflows/ci.yml:53",".github/workflows/release.yml:34",".github/workflows/release.yml:36",".github/workflows/release.yml:39",".github/workflows/release.yml:40",".github/workflows/release.yml:48","docs/releases/README.md:8","docs/releases/README.md:16","docs/releases/README.md:18","docs/releases/README.md:19","docs/releases/README.md:23","tests/test_release.py:1","tests/test_release.py:21","tests/test_release.py:23","tests/test_release.py:52","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:81","scripts/release.py:82","scripts/publish_release.py:54","scripts/verify_distribution.py:47","scripts/verify_lambda.sh:6","scripts/verify_lambda.sh:18","scripts/verify_lambda.sh:20",".github/workflows/ci.yml:29",".github/workflows/ci.yml:32",".github/workflows/ci.yml:33",".github/workflows/ci.yml:53",".github/workflows/ci.yml:54",".github/workflows/release.yml:62","docs/releases/README.md:16","docs/releases/README.md:19","docs/releases/README.md:62","docs/releases/README.md:63","docs/releases/README.md:79"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","partial: Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","partial: Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","partial: A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","name":"Python integration","label":"Python integration / Embed analysis in my application / Install a released dependency"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","label":"Ask an agent to explain un...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","acceptance":["The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit."],"guidance":["This contract covers the supplied instructions and demonstrated analysis workflow; it does not guarantee that every agent host follows the instructions.","Synthetic examples demonstrate mechanics, not real-world detection accuracy."],"path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/agents/openai.yaml","src/anomalyzer/cli.py","src/anomalyzer/mcp.py","tests/test_cli.py","tests/test_agentic.py","tests/test_mcp.py"],"automatedChecks":[{"id":"structured-cli","description":"Exercise the documented structured CLI workflow and preserve incomplete-history evidence.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence tests/test_cli.py::test_bare_json_values_cli"},{"id":"declared-relationships","description":"Check explicit ratio semantics and unavailable source evidence.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"mcp-drilldown","description":"Follow compact analysis to a finding and its source evidence.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case"}],"manualChecks":["Follow the conversion example through the CLI and MCP drilldown; compare the Skill instructions with actual evidence and check that its proposed text separates calibrated findings, early candidates, readiness, and caller-owned actions."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","criteria":[{"criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:13",".agents/skills/anomaly-detection/SKILL.md:41","tests/test_cli.py:13"],"notes":"Executable resolution and JSON commands are explicit. Manual conversion and spike CLI runs use the installed numerical engine."},{"criterion":"The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:24",".agents/skills/anomaly-detection/SKILL.md:32","tests/test_cli.py:51","tests/test_agentic.py:211"],"notes":"Instructions preserve scope, units, missingness, declared relationships, and exact alignment; existing runtime checks establish these prerequisites."},{"criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:101",".agents/skills/anomaly-detection/SKILL.md:134",".agents/skills/anomaly-detection/agents/openai.yaml:4"],"notes":"The supplied instructions lead with plain-language changes and then qualify maturity/readiness/overlap. This establishes instructional content, not measured agent compliance."},{"criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:54",".agents/skills/anomaly-detection/SKILL.md:174","src/anomalyzer/mcp.py:169","tests/test_mcp.py:54"],"notes":"The documented three-tool workflow matches executed conversion drilldown, with session-only run IDs and explicit read-only/caller-owned delivery limits."}],"ignore_findings":[],"done":true,"input_fingerprint":"751723214aa340b1227ec347c22f130749fcdbfd49327c21b4c27ef68d7f0341","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Get an evidence-grounded explanation"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","label":"See the evidence in a char...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","acceptance":["The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes."],"guidance":["Native chart rendering and interactive figure features depend on the agent host; the checked-in fallback produces a static SVG for a single dataset.","A chart explains structured analysis; it does not replace the detector result."],"path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/scripts/quick_chart.py",".agents/skills/anomaly-detection/references/visualization.md","tests/test_skill_chart.py"],"automatedChecks":[{"id":"observed-evidence","description":"Render real observed and expected data and only actual point triggers.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_preserves_real_point_evidence"},{"id":"maturity-and-resets","description":"Keep provisional evidence non-triggering and label reviewed regimes.","command":"python -m pytest -q tests/test_skill_chart.py::test_provisional_scores_are_not_charted_as_triggers tests/test_skill_chart.py::test_quick_chart_marks_reviewed_resets"},{"id":"unsupported-scope","description":"Reject a multi-dataset result rather than mixing units.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_rejects_multiple_datasets"},{"id":"output-protection","description":"Require an explicit overwrite for existing charts.","command":"python -m pytest -q tests/test_skill_chart.py::test_chart_cli_requires_explicit_overwrite"}],"manualChecks":["Inspect a spike fallback chart and the detailed figure instructions; verify that axes, initialization context, trigger markers, and interpretation limits are clear, and that the Skill requests a chart only when the user asks."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","criteria":[{"criterion":"The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:189",".agents/skills/anomaly-detection/SKILL.md:226"],"notes":"The Skill explicitly requires a visual request after analysis and separates quick versus detailed rendering instructions."},{"criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:81",".agents/skills/anomaly-detection/scripts/quick_chart.py:59","tests/test_skill_chart.py:28","tests/test_skill_chart.py:43","tests/test_skill_chart.py:54"],"notes":"Real-result tests establish observed/expected lines, trigger-only markers, provisional non-triggering, units, initialization shading, and reset boundaries; the spike chart was inspected visually."},{"criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:85",".agents/skills/anomaly-detection/scripts/quick_chart.py:251","tests/test_skill_chart.py:63","tests/test_skill_chart.py:68"],"notes":"Multiple resolved datasets raise a specific error. The CLI test preserves an existing chart and replaces it only with explicit overwrite."},{"criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/references/visualization.md:31",".agents/skills/anomaly-detection/references/visualization.md:37",".agents/skills/anomaly-detection/references/visualization.md:45",".agents/skills/anomaly-detection/references/visualization.md:86"],"notes":"Detailed instructions require index alignment, maturity distinctions, first detection timing, visible unavailable status, and qualified causal claims; native interactive rendering remains host-dependent."}],"ignore_findings":[],"done":true,"input_fingerprint":"2780b6c3e363f89a30a0709772c0a46a9a96f1567fabb27e63934a3ca80dc982","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Inspect a requested chart"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","label":"Find which segments contri...","area":"investigation","scopes":["investigation","investigation/cohorts"],"scope":"investigation/cohorts","scopeLabel":"investigation/cohorts","status":"planned","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","acceptance":["Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","Segment ranking retains absolute signed contributions and baseline support; an arithmetic contributor is not automatically labeled anomalous or causal."],"guidance":["Ratios and overlapping customer groups are not additive totals; separate segment expectations must not be assumed to sum to a fitted total expectation.","Proposed automated checks cover exact reconciliation, offsetting segment changes, missing segments, and zero total departure.","Manual review should inspect a regional decline with an offsetting increase and confirm the explanation preserves signed contributions and evidence limits."],"path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Locate contributors to a total change"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","label":"Keep reviewed outcomes bes...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","acceptance":["A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","Review export distinguishes analyst judgment from detector output; annotation alone does not change calibration, eligibility, or regime boundaries."],"guidance":["Caller-supplied reviewer identity is attribution, not built-in authentication; storage, access control, response execution, and incident ownership stay with callers.","Proposed automated checks cover evidence-version linking, unresolved reviews, revision retention, and unchanged numerical evidence after annotation.","Manual review should inspect a disputed case with two review revisions and confirm judgments, uncertainties, and original findings remain distinguishable."],"path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Record the reviewed outcome"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","label":"Find earlier reviewed case...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","acceptance":["Search is bounded to caller-selected records and explicit compatibility filters; it does not scan unrelated local files or require a model API.","Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match."],"guidance":["Similar historical evidence does not establish the same cause or authorize repeating a past response; caller-selected records may be incomplete or biased.","Proposed automated checks cover compatibility filtering, bounded corpus access, missing comparison context, and no-match behavior.","Manual review should compare superficially similar cases with different outcomes and confirm precedent is presented as investigation context."],"path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Inspect comparable reviewed history"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","label":"Change action eligibility ...","area":"investigation","scopes":["investigation","investigation/policy"],"scope":"investigation/policy","scopeLabel":"investigation/policy","status":"implemented","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","acceptance":["Default policy leaves action and notification eligibility disabled.","Explicit-rule and calibrated-departure eligibility are controlled independently.","Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged."],"guidance":["Eligibility does not send notifications, execute actions, check readiness, or confer authority to act.","Monitoring schedules, delivery, reviewed incident state, and deduplication belong to the caller."],"path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","dependencies":["investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/agent.py","src/anomalyzer/cases.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"independent-notifications","description":"Replay notification and action choices without changing numerical evidence or the original result.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_separates_action_from_notification"},{"id":"calibrated-choice","description":"Enable calibrated departures only with the corresponding policy choice.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_enables_calibrated_departures_only_when_requested"},{"id":"nontriggering-candidates","description":"Keep descriptive candidates ineligible even under a permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"}],"manualChecks":["Inspect the unit-cost example under default and allow_explicit_rules policies and compare eligibility with the unchanged numerical violation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","criteria":[{"criterion":"Default policy leaves action and notification eligibility disabled.","status":"covered","evidence":["src/anomalyzer/contracts.py:329","tests/test_capability_behaviors.py:74"],"notes":"Default flags disable both eligibility types, including after replay resets policy."},{"criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"covered","evidence":["src/anomalyzer/agent.py:47","tests/test_capability_behaviors.py:123"],"notes":"Explicit-only permission does not enable a calibrated departure; the matching calibrated choice does."},{"criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"covered","evidence":["src/anomalyzer/agent.py:53","tests/test_capability_behaviors.py:101","tests/test_capability_behaviors.py:134"],"notes":"Notification-only replay enables a met rule while keeping action disabled; permissive settings still cannot enable candidates."},{"criterion":"Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged.","status":"covered","evidence":["src/anomalyzer/agent.py:42","src/anomalyzer/cases.py:57","src/anomalyzer/cases.py:105","tests/test_capability_behaviors.py:101","tests/test_agentic.py:49"],"notes":"Assertions compare unchanged original serialization, methods, evidence references, stable IDs, and eligibility-dependent revision changes."}],"ignore_findings":[],"done":true,"input_fingerprint":"ee67bcc1ec1380b8d84a0556d838d648ae6cef3f4f9b9e34e1b71e90eb00aeae","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Review action eligibility"},"impact":0,"gaps":0,"r":38,"x":1086.6666666666667,"y":878},{"id":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","label":"Review a possible new base...","area":"investigation","scopes":["investigation","investigation/regimes"],"scope":"investigation/regimes","scopeLabel":"investigation/regimes","status":"planned","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","acceptance":["A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","Producing a proposal leaves the active regime and original findings unchanged.","A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect."],"guidance":["A proposed boundary is not a confirmed cause or a retrospectively available detection; adaptive reference updates remain a separate behavior.","Proposed automated checks cover unchanged active results, short new regimes, rejected proposals, and accepted explicit-reset equivalence.","Manual review should inspect a persistent outage-like drop and a known launch and confirm neither becomes ordinary behavior merely because it persists."],"path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Review a proposed regime transition"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","label":"Distinguish changing perfo...","area":"relationships","scopes":["relationships","relationships/cohorts"],"scope":"relationships/cohorts","scopeLabel":"relationships/cohorts","status":"planned","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","acceptance":["Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","The report preserves aggregate and cohort rates, population weights, and the declared decomposition formula and reference-period convention.","Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes."],"guidance":["Decomposition is descriptive accounting; it does not estimate an intervention effect or establish that the comparison populations are causally equivalent.","Proposed automated checks cover stable rates with changing mix, stable mix with changing rates, offsetting cohort rates, and new or empty cohorts.","Manual review should inspect a Simpson's-paradox example and confirm the aggregate movement and cohort evidence can both be understood correctly."],"path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Separate performance changes from population mix"},"impact":0,"gaps":3,"r":47,"x":1270,"y":878},{"id":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","label":"Check whether a metric fol...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","acceptance":["A supplied lag and coefficient use the predictor at the declared earlier sample position.","Derived values retain response and predictor source indexes and source cutoffs in lineage.","When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","Samples lacking the earlier predictor or awaiting coefficient training remain unavailable."],"guidance":["Lag is measured in samples and is never searched automatically.","Exactly one of coefficient or training_size is required."],"path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","dependencies":["relationships/quality/require-compatible-sources"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_examples.py","examples/agentic/lagged-response.json"],"automatedChecks":[{"id":"supplied-response","description":"Use the declared lag and coefficient with correct predictor indexes.","command":"python -m pytest -q tests/test_agentic.py::test_lagged_response_uses_supplied_coefficient_and_prior_sample"},{"id":"frozen-coefficient","description":"Keep fitted coefficient and earlier lineage invariant when future samples arrive.","command":"python -m pytest -q tests/test_agentic.py::test_estimated_lag_coefficient_is_frozen_and_prefix_invariant"},{"id":"delayed-example","description":"Run the checked-in lagged-response demonstration against expected results.","command":"python -m pytest -q tests/test_examples.py::test_agentic_demo_checks_all_cases"},{"id":"pending-training","description":"Retain unavailable lineage until the full declared training prefix arrives, then freeze its coefficient.","command":"python -m pytest -q tests/test_agentic.py::test_lag_fit_waits_for_complete_training_prefix"}],"manualChecks":["Inspect the lagged-response example's leading unavailable values, coefficient parameters, and first usable derived sample."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","criteria":[{"criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"covered","evidence":["src/anomalyzer/relationships.py:115","tests/test_agentic.py:254"],"notes":"Supplied lag/coefficient uses the declared earlier predictor position and preserves the expected response residual."},{"criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"covered","evidence":["src/anomalyzer/relationships.py:152","tests/test_agentic.py:254","examples/agentic/lagged-response.json"],"notes":"Available and pending lineage retains existing source indexes and timestamp cutoffs; missing predictors do not become zero."},{"criterion":"When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","status":"covered","evidence":["src/anomalyzer/relationships.py:47","tests/test_agentic.py:294","tests/test_agentic.py:330"],"notes":"The coefficient is fitted once after the complete declared prefix; the independent 20/7 result and prefix-invariance assertions pass."},{"criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"covered","evidence":["src/anomalyzer/relationships.py:122","tests/test_agentic.py:330"],"notes":"Short prefixes retain null derived values and explicit unavailable reasons, while an unidentifiable completed fit has a visible error."}],"ignore_findings":[],"done":true,"input_fingerprint":"fc4516ef6ab903dfa4918268d97049aff04fbf9556620ccb8259ff830ba85785","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Check delayed responses"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878},{"id":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","label":"Inspect related metrics be...","area":"relationships","scopes":["relationships","relationships/resolution"],"scope":"relationships/resolution","scopeLabel":"relationships/resolution","status":"implemented","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","acceptance":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"guidance":["Daily ratios and differences are supported. Daily lagged, normalized-residual and joint-condition definitions remain explicitly inapplicable; their native evidence remains eligible. Local-calendar aggregation is separate work.","Initial support should use explicit relationships and compatible boundaries; cross-cadence joining or automatic resampling needs a separate contract.","Proposed automated checks cover ratio-of-totals arithmetic, missing sources, mismatched period boundaries, and open-period evidence separation.","Manual review should inspect a conversion decline within an unfinished day and verify the explanation separates native evidence from partial totals."],"path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","dependencies":["data/quality/analyze-sparse-observations","relationships/comparisons/compare-declared-metrics"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","src/anomalyzer/relationships.py","src/anomalyzer/agent.py","tests/test_sparse.py","examples/sparse/unfinished_daily.json","docs/sparse-data.md"],"automatedChecks":[{"id":"ratio-of-sums","description":"Compute daily ratios from compatible sums and preserve missing-child coverage.","command":"python -m pytest -q tests/test_sparse.py::test_complete_daily_ratio_divides_sums_and_missing_child_invalidates_only_its_view"},{"id":"view-policies","description":"Keep native/daily rules separate and preserve case identity on replay.","command":"python -m pytest -q tests/test_sparse.py::test_daily_rules_are_not_silently_reused_from_hourly_rules tests/test_sparse.py::test_multi_resolution_mcp_drilldown_selects_correct_view_and_replays_case_ids"},{"id":"view-limits","description":"Report unsupported daily lag semantics without discarding native evidence.","command":"python -m pytest -q tests/test_sparse.py::test_unsupported_daily_relationship_is_explicit_and_does_not_reinterpret_lag"}],"manualChecks":["Inspect progression chapter 22 and confirm incomplete progress is not presented as a completed total or independent corroboration."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for relationships/resolution/compare-finalized-subperiods.","criteria":[{"criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:22","src/anomalyzer/sparse_views.py:139","src/anomalyzer/sparse_views.py:143","src/anomalyzer/sparse_views.py:164","src/anomalyzer/agent.py:14","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:23","src/anomalyzer/agent.py:29","tests/test_sparse.py:18","tests/test_sparse.py:59","tests/test_sparse.py:87","tests/test_sparse.py:99","tests/test_sparse.py:117","docs/sparse-data.md:31","docs/sparse-data.md:127","docs/sparse-data.md:159","docs/sparse-data.md:199","docs/sparse-data.md:203"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:122","src/anomalyzer/sparse_views.py:123","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:36","src/anomalyzer/agent.py:34","tests/test_sparse.py:10","tests/test_sparse.py:40","tests/test_sparse.py:50","tests/test_sparse.py:60","tests/test_sparse.py:62","docs/sparse-data.md:17","docs/sparse-data.md:41","docs/sparse-data.md:68","docs/sparse-data.md:97","docs/sparse-data.md:99"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:70","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:170","src/anomalyzer/relationships.py:90","src/anomalyzer/relationships.py:173","tests/test_sparse.py:230","tests/test_sparse.py:259","tests/test_sparse.py:313","docs/sparse-data.md:41","docs/sparse-data.md:111","docs/sparse-data.md:194","docs/sparse-data.md:197","docs/sparse-data.md:198"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:130","src/anomalyzer/sparse_views.py:159","src/anomalyzer/relationships.py:157","src/anomalyzer/relationships.py:192","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:47","tests/test_sparse.py:217","tests/test_sparse.py:247","tests/test_sparse.py:251","tests/test_sparse.py:264","tests/test_sparse.py:386","examples/sparse/unfinished_daily.json:208","docs/sparse-data.md:8","docs/sparse-data.md:11","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:45"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","partial: Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","partial: Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","partial: Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","name":"Related-metric analysis","label":"Related-metric analysis / Monitor the periods my business uses / Compare related metrics while the period is open"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878}],"links":[{"source":"data/inputs/accept-analysis-requests","target":"analysis/evidence/calibrate-departures","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/cli/analyze-files-and-streams","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/identify-residual-patterns","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"analysis/expectations/model-seasonal-level-and-growth","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"interfaces/python/analyze-through-public-api","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/expectations/reset-declared-regimes","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"investigation/assessments/describe-early-changes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"relationships/quality/require-compatible-sources","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"data/quality/exclude-incomplete-periods","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/mcp/connect-read-only-local-server","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/rules/check-declared-numerical-rules","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/report-point-anomaly-episodes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/evidence/assess-reference-readiness","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/adapt-to-changing-growth","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/protect-reference-from-extremes","kind":"committed"},{"source":"data/quality/exclude-incomplete-periods","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"execution/resilience/retain-evidence-under-failure-and-limits","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/mcp/connect-read-only-local-server","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/visualize-analysis-on-request","kind":"committed"},{"source":"investigation/rules/check-declared-numerical-rules","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/evaluate-delayed-response","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"},{"source":"relationships/comparisons/compare-declared-metrics","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"}]};
 const width = 1440;
 const height = 980;
 const padding = 74;
diff --git a/.capabilities/dependency-preview.svg b/.capabilities/dependency-preview.svg
index adf0e31..6672b44 100644
--- a/.capabilities/dependency-preview.svg
+++ b/.capabilities/dependency-preview.svg
@@ -81,50 +81,64 @@
   </g>
   <g id="graph">
     <g id="links">
-      <path class="link committed" d="M726.7,209.9 Q538.3,179.7 406.3,317.4" />
-      <path class="link committed" d="M941.8,225.6 Q906.2,230.9 925.7,261.1" />
-      <path class="link committed" d="M722.8,183.0 Q606.9,110.6 485.5,173.5" />
-      <path class="link committed" d="M872.5,179.4 Q890.4,208.4 902.8,176.7" />
-      <path class="link committed" d="M1203.1,236.2 Q1215.2,211.3 1188.8,219.5" />
-      <path class="link committed" d="M337.7,166.3 Q709.8,258.5 1083.0,170.6" />
-      <path class="link committed" d="M311.4,270.2 Q336.7,237.8 298.4,222.9" />
-      <path class="link committed" d="M1005.2,216.9 Q998.5,257.0 1039.1,257.3" />
-      <path class="link committed" d="M1118.9,224.0 Q1088.2,224.6 1104.8,250.3" />
-      <path class="link committed" d="M871.2,201.3 Q1009.6,313.0 1181.2,265.9" />
-      <path class="link committed" d="M279.2,624.8 Q808.0,489.6 1264.0,189.8" />
-      <path class="link committed" d="M361.6,169.0 Q268.1,111.1 171.0,162.8" />
-      <path class="link committed" d="M274.1,294.1 Q248.8,206.4 158.2,195.9" />
-      <path class="link committed" d="M723.5,175.8 Q700.8,139.3 669.1,168.3" />
-      <path class="link committed" d="M167.0,161.0 Q364.3,250.7 561.6,161.0" />
-      <path class="link committed" d="M337.7,165.2 Q450.5,223.4 561.6,161.8" />
-      <path class="link committed" d="M273.5,614.9 Q512.8,553.5 646.9,346.0" />
-      <path class="link committed" d="M1063.2,355.6 Q1028.9,372.9 1050.3,404.7" />
-      <path class="link committed" d="M1219.3,329.3 Q1179.7,346.9 1197.4,386.5" />
-      <path class="link committed" d="M1319.2,221.0 Q1277.6,286.5 1321.9,350.2" />
-      <path class="link committed" d="M406.2,347.2 Q832.1,463.8 1269.1,400.8" />
-      <path class="link committed" d="M302.5,411.7 Q133.6,582.6 130.9,822.8" />
-      <path class="link committed" d="M264.3,308.9 Q259.2,273.1 227.5,290.4" />
-      <path class="link committed" d="M279.6,396.8 Q171.3,427.6 145.8,537.2" />
-      <path class="link committed" d="M466.4,210.8 Q464.6,245.9 499.2,239.8" />
-      <path class="link committed" d="M644.4,333.0 Q426.3,324.1 274.9,481.2" />
-      <path class="link committed" d="M391.2,222.8 Q263.6,305.4 254.2,457.1" />
-      <path class="link committed" d="M819.7,257.7 Q793.4,362.6 873.7,435.0" />
-      <path class="link committed" d="M914.7,386.6 Q888.6,479.4 959.2,545.2" />
-      <path class="link committed" d="M1124.1,226.4 Q981.8,355.3 991.2,547.1" />
-      <path class="link committed" d="M934.2,221.5 Q720.1,388.4 657.6,652.5" />
-      <path class="link committed" d="M859.2,376.7 Q692.5,467.2 659.8,654.0" />
-      <path class="link committed" d="M993.5,482.2 Q790.8,502.8 674.3,669.9" />
-      <path class="link committed" d="M897.9,389.0 Q828.5,529.8 901.1,669.0" />
-      <path class="link committed" d="M1055.7,353.1 Q913.4,481.4 920.7,672.9" />
-      <path class="link committed" d="M875.7,385.3 Q743.7,517.6 764.9,703.2" />
-      <path class="link committed" d="M1176.6,486.9 Q1094.0,650.6 1166.6,819.0" />
-      <path class="link committed" d="M1148.6,231.0 Q1066.2,527.3 1163.8,819.0" />
-      <path class="link committed" d="M1293.5,215.2 Q1098.4,405.4 1064.5,675.7" />
-      <path class="link committed" d="M651.2,196.0 Q844.1,511.7 1163.1,699.0" />
-      <path class="link committed" d="M1305.3,450.9 Q1201.1,546.7 1215.5,687.5" />
+      <path class="link committed" d="M643.0,307.1 Q452.1,168.4 218.8,204.6" />
+      <path class="link committed" d="M943.4,220.6 Q902.2,223.4 914.6,262.8" />
+      <path class="link committed" d="M203.3,229.2 Q234.6,315.0 325.9,315.7" />
+      <path class="link committed" d="M646.4,296.5 Q545.0,174.9 387.8,194.5" />
+      <path class="link committed" d="M789.3,284.5 Q883.7,290.5 922.4,204.1" />
+      <path class="link committed" d="M1277.8,292.6 Q1264.5,222.9 1194.1,214.6" />
+      <path class="link committed" d="M436.6,338.8 Q781.4,349.8 1084.9,185.7" />
+      <path class="link committed" d="M660.9,379.0 Q438.6,455.9 335.6,667.5" />
+      <path class="link committed" d="M141.7,443.0 Q146.7,580.6 262.3,655.4" />
+      <path class="link committed" d="M363.9,410.8 Q271.0,514.5 305.5,649.4" />
+      <path class="link committed" d="M999.6,226.7 Q980.2,248.5 1009.3,250.9" />
+      <path class="link committed" d="M1108.1,216.5 Q1064.7,217.8 1069.8,260.9" />
+      <path class="link committed" d="M800.8,326.5 Q1029.3,418.1 1259.0,329.7" />
+      <path class="link committed" d="M859.4,166.0 Q1059.2,256.0 1259.0,166.0" />
+      <path class="link committed" d="M287.5,215.1 Q179.5,249.4 153.7,359.8" />
+      <path class="link committed" d="M133.3,260.5 Q89.8,298.7 123.3,345.9" />
+      <path class="link committed" d="M749.4,164.3 Q724.9,134.7 698.5,162.7" />
+      <path class="link committed" d="M988.7,340.4 Q894.9,354.0 862.9,443.2" />
+      <path class="link committed" d="M1272.8,361.5 Q1204.5,360.6 1181.4,425.0" />
+      <path class="link committed" d="M1280.0,205.7 Q1220.5,214.7 1214.1,274.4" />
+      <path class="link committed" d="M216.4,195.2 Q658.8,340.3 1123.2,306.8" />
+      <path class="link committed" d="M138.7,260.9 Q36.5,536.0 114.2,819.0" />
+      <path class="link committed" d="M216.7,179.3 Q326.8,229.7 426.0,160.4" />
+      <path class="link committed" d="M159.7,258.9 Q120.9,424.2 231.4,553.3" />
+      <path class="link committed" d="M365.0,218.4 Q385.6,341.4 499.7,392.0" />
+      <path class="link committed" d="M604.1,190.4 Q390.9,235.0 282.9,424.1" />
+      <path class="link committed" d="M310.9,228.4 Q233.4,302.0 257.4,406.1" />
+      <path class="link committed" d="M672.3,389.5 Q570.3,429.5 558.8,538.4" />
+      <path class="link committed" d="M339.8,680.4 Q436.2,683.4 486.1,600.8" />
+      <path class="link committed" d="M411.6,402.4 Q416.2,496.7 501.8,536.7" />
+      <path class="link committed" d="M730.5,405.3 Q688.6,490.7 749.7,563.6" />
+      <path class="link committed" d="M914.5,372.0 Q907.1,414.2 949.8,419.4" />
+      <path class="link committed" d="M1116.9,222.8 Q1008.5,289.3 1003.7,416.4" />
+      <path class="link committed" d="M1006.2,223.5 Q1019.9,426.2 1184.1,545.9" />
+      <path class="link committed" d="M926.3,360.7 Q999.6,519.1 1169.7,557.9" />
+      <path class="link committed" d="M869.1,491.2 Q994.8,602.8 1160.7,575.2" />
+      <path class="link committed" d="M867.6,384.3 Q770.7,531.9 827.8,699.0" />
+      <path class="link committed" d="M1007.8,355.3 Q845.7,491.6 842.7,703.4" />
+      <path class="link committed" d="M880.9,384.6 Q836.3,474.1 894.3,555.6" />
+      <path class="link committed" d="M1124.2,504.1 Q1008.0,642.7 1040.2,820.6" />
+      <path class="link committed" d="M1137.1,230.1 Q997.8,509.7 1036.0,819.7" />
+      <path class="link committed" d="M1288.6,212.5 Q1109.7,329.2 1080.6,540.9" />
+      <path class="link committed" d="M346.5,706.9 Q717.8,798.2 1089.7,709.6" />
+      <path class="link committed" d="M1172.0,363.2 Q1086.4,505.2 1143.4,661.0" />
     </g>
     <g id="nodes">
-      <g class="node partial" data-capability="analysis/evidence/calibrate-departures" transform="translate(331.4 342.5)">
+      <g class="node partial" data-capability="data/inputs/accept-analysis-requests" transform="translate(720.8 325.9)">
+        <title>Bring my metrics into an analysis</title>
+        <circle class="node-halo" r="81" />
+        <circle class="node-ring" r="73" />
+        <circle class="node-core" r="63" />
+        <text class="label" y="-3">
+          <tspan x="0" dy="-3">Bring my metrics</tspan>
+          <tspan x="0" dy="14">into an a...</tspan>
+        </text>
+        <text class="meta" y="61">data/inputs</text>
+      </g>
+      <g class="node partial" data-capability="analysis/evidence/calibrate-departures" transform="translate(142.0 186.0)">
         <title>Understand how much history supports a finding</title>
         <circle class="node-halo" r="76" />
         <circle class="node-ring" r="68" />
@@ -135,18 +149,7 @@
         </text>
         <text class="meta" y="56">analysis/evidence</text>
       </g>
-      <g class="node partial" data-capability="data/inputs/accept-analysis-requests" transform="translate(797.8 186.0)">
-        <title>Bring my metrics into an analysis</title>
-        <circle class="node-halo" r="76" />
-        <circle class="node-ring" r="68" />
-        <circle class="node-core" r="58" />
-        <text class="label" y="-3">
-          <tspan x="0" dy="-3">Bring my metrics</tspan>
-          <tspan x="0" dy="14">into an a...</tspan>
-        </text>
-        <text class="meta" y="56">data/inputs</text>
-      </g>
-      <g class="node partial" data-capability="interfaces/cli/analyze-files-and-streams" transform="translate(897.1 324.0)">
+      <g class="node partial" data-capability="interfaces/cli/analyze-files-and-streams" transform="translate(875.8 319.8)">
         <title>Analyze local files from the command line</title>
         <circle class="node-halo" r="66" />
         <circle class="node-ring" r="58" />
@@ -157,7 +160,18 @@
         </text>
         <text class="meta" y="46">interfaces/cli</text>
       </g>
-      <g class="node partial" data-capability="analysis/expectations/model-seasonal-level-and-growth" transform="translate(421.5 171.0)">
+      <g class="node partial" data-capability="analysis/detection/identify-residual-patterns" transform="translate(378.2 352.5)">
+        <title>Spot sustained shifts and changing patterns</title>
+        <circle class="node-halo" r="61" />
+        <circle class="node-ring" r="53" />
+        <circle class="node-core" r="43" />
+        <text class="label" y="-3">
+          <tspan x="0" dy="-3">Spot sustained shifts</tspan>
+          <tspan x="0" dy="14">and ...</tspan>
+        </text>
+        <text class="meta" y="41">analysis/detection</text>
+      </g>
+      <g class="node partial" data-capability="analysis/expectations/model-seasonal-level-and-growth" transform="translate(328.3 171.0)">
         <title>Account for ordinary seasonal patterns and growth</title>
         <circle class="node-halo" r="61" />
         <circle class="node-ring" r="53" />
@@ -168,7 +182,7 @@
         </text>
         <text class="meta" y="41">analysis/expectations</text>
       </g>
-      <g class="node partial" data-capability="interfaces/python/analyze-through-public-api" transform="translate(966.6 171.0)">
+      <g class="node partial" data-capability="interfaces/python/analyze-through-public-api" transform="translate(977.2 171.0)">
         <title>Embed local analysis in my application</title>
         <circle class="node-halo" r="61" />
         <circle class="node-ring" r="53" />
@@ -179,7 +193,7 @@
         </text>
         <text class="meta" y="41">interfaces/python</text>
       </g>
-      <g class="node partial" data-capability="investigation/cases/group-related-evidence" transform="translate(1147.0 171.0)">
+      <g class="node partial" data-capability="investigation/cases/group-related-evidence" transform="translate(1147.2 171.0)">
         <title>Investigate related findings together</title>
         <circle class="node-halo" r="61" />
         <circle class="node-ring" r="53" />
@@ -190,18 +204,18 @@
         </text>
         <text class="meta" y="41">investigation/cases</text>
       </g>
-      <g class="node partial" data-capability="analysis/detection/identify-residual-patterns" transform="translate(282.7 166.0)">
-        <title>Spot sustained shifts and changing patterns</title>
+      <g class="node partial" data-capability="data/quality/analyze-sparse-observations" transform="translate(291.5 706.7)">
+        <title>Assess sparse observations without inventing consumption</title>
         <circle class="node-halo" r="56" />
         <circle class="node-ring" r="48" />
         <circle class="node-core" r="38" />
         <text class="label" y="-3">
-          <tspan x="0" dy="-3">Spot sustained shifts</tspan>
-          <tspan x="0" dy="14">and ...</tspan>
+          <tspan x="0" dy="-3">Assess sparse</tspan>
+          <tspan x="0" dy="14">observations...</tspan>
         </text>
-        <text class="meta" y="36">analysis/detection</text>
+        <text class="meta" y="36">data/quality</text>
       </g>
-      <g class="node partial" data-capability="data/quality/prepare-timestamped-series" transform="translate(228.9 647.0)">
+      <g class="node partial" data-capability="data/quality/prepare-timestamped-series" transform="translate(804.4 166.0)">
         <title>Analyze observations in the right time order</title>
         <circle class="node-halo" r="56" />
         <circle class="node-ring" r="48" />
@@ -212,7 +226,7 @@
         </text>
         <text class="meta" y="36">data/quality</text>
       </g>
-      <g class="node partial" data-capability="interfaces/mcp/investigate-findings-and-evidence" transform="translate(1077.1 302.4)">
+      <g class="node partial" data-capability="interfaces/mcp/investigate-findings-and-evidence" transform="translate(1031.3 305.6)">
         <title>Let an agent inspect the evidence behind a finding</title>
         <circle class="node-halo" r="56" />
         <circle class="node-ring" r="48" />
@@ -223,7 +237,7 @@
         </text>
         <text class="meta" y="36">interfaces/mcp</text>
       </g>
-      <g class="node partial" data-capability="investigation/assessments/describe-early-changes" transform="translate(1238.9 277.9)">
+      <g class="node partial" data-capability="investigation/assessments/describe-early-changes" transform="translate(1318.0 330.1)">
         <title>Spot early changes with a short history</title>
         <circle class="node-halo" r="56" />
         <circle class="node-ring" r="48" />
@@ -245,7 +259,7 @@
         </text>
         <text class="meta" y="36">relationships/quality</text>
       </g>
-      <g class="node partial" data-capability="analysis/expectations/reset-declared-regimes" transform="translate(117.0 161.0)">
+      <g class="node partial" data-capability="analysis/expectations/reset-declared-regimes" transform="translate(117.0 399.5)">
         <title>Compare against a new baseline after a known change</title>
         <circle class="node-halo" r="51" />
         <circle class="node-ring" r="43" />
@@ -256,18 +270,7 @@
         </text>
         <text class="meta" y="31">analysis/expectations</text>
       </g>
-      <g class="node partial" data-capability="data/quality/analyze-sparse-observations" transform="translate(615.6 161.0)">
-        <title>Assess sparse observations without inventing consumption</title>
-        <circle class="node-halo" r="51" />
-        <circle class="node-ring" r="43" />
-        <circle class="node-core" r="33" />
-        <text class="label" y="-3">
-          <tspan x="0" dy="-3">Assess sparse</tspan>
-          <tspan x="0" dy="14">observations...</tspan>
-        </text>
-        <text class="meta" y="31">data/quality</text>
-      </g>
-      <g class="node partial" data-capability="data/quality/exclude-incomplete-periods" transform="translate(690.8 314.4)">
+      <g class="node partial" data-capability="data/quality/exclude-incomplete-periods" transform="translate(644.5 161.0)">
         <title>Avoid mistaking unfinished periods for drops</title>
         <circle class="node-halo" r="51" />
         <circle class="node-ring" r="43" />
@@ -278,7 +281,7 @@
         </text>
         <text class="meta" y="31">data/quality</text>
       </g>
-      <g class="node partial" data-capability="interfaces/mcp/connect-read-only-local-server" transform="translate(1036.6 456.9)">
+      <g class="node partial" data-capability="interfaces/mcp/connect-read-only-local-server" transform="translate(821.1 477.4)">
         <title>Give an agent read-only access to local analysis</title>
         <circle class="node-halo" r="51" />
         <circle class="node-ring" r="43" />
@@ -289,7 +292,7 @@
         </text>
         <text class="meta" y="31">interfaces/mcp</text>
       </g>
-      <g class="node partial" data-capability="investigation/rules/check-declared-numerical-rules" transform="translate(1178.1 436.9)">
+      <g class="node partial" data-capability="investigation/rules/check-declared-numerical-rules" transform="translate(1137.0 455.8)">
         <title>Find values that break my stated limits</title>
         <circle class="node-halo" r="51" />
         <circle class="node-ring" r="43" />
@@ -300,7 +303,7 @@
         </text>
         <text class="meta" y="31">investigation/rules</text>
       </g>
-      <g class="node partial" data-capability="relationships/comparisons/compare-declared-metrics" transform="translate(1323.0 404.2)">
+      <g class="node partial" data-capability="relationships/comparisons/compare-declared-metrics" transform="translate(1176.8 313.4)">
         <title>Find changes in conversion, unit cost, and metric gaps</title>
         <circle class="node-halo" r="51" />
         <circle class="node-ring" r="43" />
@@ -322,7 +325,7 @@
         </text>
         <text class="meta" y="26">analysis/detection</text>
       </g>
-      <g class="node partial" data-capability="analysis/evidence/assess-reference-readiness" transform="translate(183.7 268.4)">
+      <g class="node partial" data-capability="analysis/evidence/assess-reference-readiness" transform="translate(474.8 156.0)">
         <title>Know when the expected baseline is weak</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -333,7 +336,7 @@
         </text>
         <text class="meta" y="26">analysis/evidence</text>
       </g>
-      <g class="node partial" data-capability="analysis/expectations/adapt-to-changing-growth" transform="translate(112.0 572.7)">
+      <g class="node partial" data-capability="analysis/expectations/adapt-to-changing-growth" transform="translate(243.0 600.9)">
         <title>Follow evolving growth while exposing unusual values</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -344,7 +347,7 @@
         </text>
         <text class="meta" y="26">analysis/expectations</text>
       </g>
-      <g class="node uncovered" data-capability="analysis/expectations/model-multiple-seasonal-cycles" transform="translate(121.0 403.2)">
+      <g class="node uncovered" data-capability="analysis/expectations/model-multiple-seasonal-cycles" transform="translate(234.8 287.4)">
         <title>Account for daily and weekly patterns together</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -355,7 +358,7 @@
         </text>
         <text class="meta" y="35">analysis/expectations</text>
       </g>
-      <g class="node partial" data-capability="analysis/expectations/protect-reference-from-extremes" transform="translate(535.8 272.3)">
+      <g class="node partial" data-capability="analysis/expectations/protect-reference-from-extremes" transform="translate(529.8 430.7)">
         <title>Limit how extreme values distort later comparisons</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -366,7 +369,7 @@
         </text>
         <text class="meta" y="26">analysis/expectations</text>
       </g>
-      <g class="node partial" data-capability="analysis/expectations/select-season-length" transform="translate(112.0 724.9)">
+      <g class="node partial" data-capability="analysis/expectations/select-season-length" transform="translate(112.0 554.2)">
         <title>Choose the repeating cycle for a fair comparison</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -377,7 +380,7 @@
         </text>
         <text class="meta" y="26">analysis/expectations</text>
       </g>
-      <g class="node uncovered" data-capability="analysis/rates/assess-bounded-count-rates" transform="translate(370.3 531.3)">
+      <g class="node uncovered" data-capability="analysis/rates/assess-bounded-count-rates" transform="translate(538.1 286.3)">
         <title>Interpret rate changes with their supporting volume</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -388,7 +391,7 @@
         </text>
         <text class="meta" y="35">analysis/rates</text>
       </g>
-      <g class="node partial" data-capability="analysis/resolution/monitor-within-unfinished-days" transform="translate(229.4 499.4)">
+      <g class="node partial" data-capability="analysis/resolution/monitor-within-unfinished-days" transform="translate(243.3 453.0)">
         <title>See changes before the day is complete</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -399,7 +402,7 @@
         </text>
         <text class="meta" y="26">analysis/resolution</text>
       </g>
-      <g class="node uncovered" data-capability="data/calendar/aggregate-local-calendar-periods" transform="translate(393.9 859.0)">
+      <g class="node uncovered" data-capability="data/calendar/aggregate-local-calendar-periods" transform="translate(393.4 859.0)">
         <title>Compare days and weeks in my business timezone</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -410,7 +413,7 @@
         </text>
         <text class="meta" y="35">data/calendar</text>
       </g>
-      <g class="node uncovered" data-capability="data/cohorts/analyze-declared-entity-batches" transform="translate(512.5 418.6)">
+      <g class="node uncovered" data-capability="data/cohorts/analyze-declared-entity-batches" transform="translate(384.8 529.1)">
         <title>Analyze many entities without mixing their histories</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -421,7 +424,18 @@
         </text>
         <text class="meta" y="35">data/cohorts</text>
       </g>
-      <g class="node uncovered" data-capability="data/quality/report-source-freshness" transform="translate(534.9 578.6)">
+      <g class="node partial" data-capability="data/quality/identify-distinct-events" transform="translate(529.1 577.4)">
+        <title>Inspect distinct events sharing an original timestamp</title>
+        <circle class="node-halo" r="46" />
+        <circle class="node-ring" r="38" />
+        <circle class="node-core" r="28" />
+        <text class="label" y="-3">
+          <tspan x="0" dy="-3">Inspect distinct</tspan>
+          <tspan x="0" dy="14">events sh...</tspan>
+        </text>
+        <text class="meta" y="26">data/quality</text>
+      </g>
+      <g class="node uncovered" data-capability="data/quality/report-source-freshness" transform="translate(122.2 716.7)">
         <title>Know when missing updates weaken the conclusion</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -432,7 +446,7 @@
         </text>
         <text class="meta" y="35">data/quality</text>
       </g>
-      <g class="node uncovered" data-capability="data/revisions/reconcile-late-and-corrected-observations" transform="translate(237.1 818.9)">
+      <g class="node uncovered" data-capability="data/revisions/reconcile-late-and-corrected-observations" transform="translate(238.9 859.0)">
         <title>See how late or corrected data changes the evidence</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -443,7 +457,7 @@
         </text>
         <text class="meta" y="35">data/revisions</text>
       </g>
-      <g class="node uncovered" data-capability="evaluation/comparisons/compare-monitor-configurations" transform="translate(671.0 476.4)">
+      <g class="node uncovered" data-capability="evaluation/comparisons/compare-monitor-configurations" transform="translate(660.0 508.9)">
         <title>Choose settings using held-out historical evidence</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -454,7 +468,7 @@
         </text>
         <text class="meta" y="35">evaluation/comparisons</text>
       </g>
-      <g class="node uncovered" data-capability="evaluation/outcomes/compare-findings-with-reviewed-incidents" transform="translate(386.9 687.7)">
+      <g class="node uncovered" data-capability="evaluation/outcomes/compare-findings-with-reviewed-incidents" transform="translate(456.0 730.8)">
         <title>Measure detection timing against reviewed incidents</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -465,7 +479,7 @@
         </text>
         <text class="meta" y="35">evaluation/outcomes</text>
       </g>
-      <g class="node uncovered" data-capability="evaluation/replay/replay-historical-observations" transform="translate(509.2 780.2)">
+      <g class="node uncovered" data-capability="evaluation/replay/replay-historical-observations" transform="translate(626.2 703.3)">
         <title>Replay history as it would have arrived</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -476,7 +490,7 @@
         </text>
         <text class="meta" y="35">evaluation/replay</text>
       </g>
-      <g class="node uncovered" data-capability="execution/incremental/append-finalized-observations" transform="translate(666.0 859.0)">
+      <g class="node uncovered" data-capability="execution/incremental/append-finalized-observations" transform="translate(877.3 859.0)">
         <title>Update evidence when the next observations arrive</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -487,7 +501,7 @@
         </text>
         <text class="meta" y="35">execution/incremental</text>
       </g>
-      <g class="node partial" data-capability="execution/resilience/retain-evidence-under-failure-and-limits" transform="translate(888.0 481.8)">
+      <g class="node partial" data-capability="execution/resilience/retain-evidence-under-failure-and-limits" transform="translate(755.7 612.2)">
         <title>Keep useful evidence when analysis cannot finish</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -498,7 +512,7 @@
         </text>
         <text class="meta" y="26">execution/resilience</text>
       </g>
-      <g class="node uncovered" data-capability="execution/state/save-and-restore-analysis-state" transform="translate(792.0 591.0)">
+      <g class="node uncovered" data-capability="execution/state/save-and-restore-analysis-state" transform="translate(551.6 859.0)">
         <title>Resume local analysis after a restart</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -509,7 +523,7 @@
         </text>
         <text class="meta" y="35">execution/state</text>
       </g>
-      <g class="node partial" data-capability="interfaces/cli/explain-findings-in-readable-output" transform="translate(972.4 592.4)">
+      <g class="node partial" data-capability="interfaces/cli/explain-findings-in-readable-output" transform="translate(979.0 458.7)">
         <title>Understand findings in a readable summary</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -520,7 +534,7 @@
         </text>
         <text class="meta" y="26">interfaces/cli</text>
       </g>
-      <g class="node uncovered" data-capability="interfaces/export/package-investigation-evidence" transform="translate(1132.5 591.2)">
+      <g class="node uncovered" data-capability="interfaces/export/package-investigation-evidence" transform="translate(719.6 859.0)">
         <title>Share an investigation that can be inspected later</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -531,18 +545,18 @@
         </text>
         <text class="meta" y="35">interfaces/export</text>
       </g>
-      <g class="node partial" data-capability="interfaces/python/install-versioned-library" transform="translate(629.5 696.3)">
+      <g class="node partial" data-capability="interfaces/python/install-versioned-library" transform="translate(1207.8 588.8)">
         <title>Install a versioned analysis library</title>
-        <circle class="node-halo" r="49" />
-        <circle class="node-ring" r="41" />
-        <circle class="node-core" r="31" />
+        <circle class="node-halo" r="46" />
+        <circle class="node-ring" r="38" />
+        <circle class="node-core" r="28" />
         <text class="label" y="-3">
           <tspan x="0" dy="-3">Install a</tspan>
           <tspan x="0" dy="14">versioned analys...</tspan>
         </text>
-        <text class="meta" y="29">interfaces/python</text>
+        <text class="meta" y="26">interfaces/python</text>
       </g>
-      <g class="node partial" data-capability="interfaces/skill/guide-anomaly-investigation" transform="translate(901.6 718.0)">
+      <g class="node partial" data-capability="interfaces/skill/guide-anomaly-investigation" transform="translate(821.7 747.6)">
         <title>Ask an agent to explain unusual changes in my metrics</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -553,7 +567,7 @@
         </text>
         <text class="meta" y="26">interfaces/skill</text>
       </g>
-      <g class="node partial" data-capability="interfaces/skill/visualize-analysis-on-request" transform="translate(748.8 749.5)">
+      <g class="node partial" data-capability="interfaces/skill/visualize-analysis-on-request" transform="translate(898.1 604.4)">
         <title>See the evidence in a chart when I ask for one</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -564,7 +578,7 @@
         </text>
         <text class="meta" y="26">interfaces/skill</text>
       </g>
-      <g class="node uncovered" data-capability="investigation/cohorts/explain-additive-segment-contributions" transform="translate(1319.0 859.0)">
+      <g class="node uncovered" data-capability="investigation/cohorts/explain-additive-segment-contributions" transform="translate(1319.0 495.8)">
         <title>Find which segments contribute to an aggregate change</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -575,7 +589,7 @@
         </text>
         <text class="meta" y="35">investigation/cohorts</text>
       </g>
-      <g class="node uncovered" data-capability="investigation/history/record-reviewed-case-outcomes" transform="translate(1299.6 546.6)">
+      <g class="node uncovered" data-capability="investigation/history/record-reviewed-case-outcomes" transform="translate(1168.6 859.0)">
         <title>Keep reviewed outcomes beside the original evidence</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -586,7 +600,7 @@
         </text>
         <text class="meta" y="35">investigation/history</text>
       </g>
-      <g class="node uncovered" data-capability="investigation/history/retrieve-comparable-reviewed-cases" transform="translate(835.8 859.0)">
+      <g class="node uncovered" data-capability="investigation/history/retrieve-comparable-reviewed-cases" transform="translate(1319.0 691.6)">
         <title>Find earlier reviewed cases with comparable evidence</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -597,7 +611,7 @@
         </text>
         <text class="meta" y="35">investigation/history</text>
       </g>
-      <g class="node partial" data-capability="investigation/policy/replay-action-eligibility" transform="translate(1165.1 868.0)">
+      <g class="node partial" data-capability="investigation/policy/replay-action-eligibility" transform="translate(1027.7 868.0)">
         <title>Change action eligibility without rerunning analysis</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -608,7 +622,7 @@
         </text>
         <text class="meta" y="26">investigation/policy</text>
       </g>
-      <g class="node uncovered" data-capability="investigation/regimes/propose-reviewed-baseline-changes" transform="translate(996.5 859.0)">
+      <g class="node uncovered" data-capability="investigation/regimes/propose-reviewed-baseline-changes" transform="translate(974.3 735.4)">
         <title>Review a possible new baseline before accepting it</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -619,7 +633,7 @@
         </text>
         <text class="meta" y="35">investigation/regimes</text>
       </g>
-      <g class="node uncovered" data-capability="relationships/cohorts/separate-rate-and-population-mix" transform="translate(1319.0 723.2)">
+      <g class="node uncovered" data-capability="relationships/cohorts/separate-rate-and-population-mix" transform="translate(1319.0 859.0)">
         <title>Distinguish changing performance from a changing population mix</title>
         <circle class="node-halo" r="55" />
         <circle class="node-ring" r="47" />
@@ -630,7 +644,7 @@
         </text>
         <text class="meta" y="35">relationships/cohorts</text>
       </g>
-      <g class="node partial" data-capability="relationships/comparisons/evaluate-delayed-response" transform="translate(1042.7 719.6)">
+      <g class="node partial" data-capability="relationships/comparisons/evaluate-delayed-response" transform="translate(1054.4 582.3)">
         <title>Check whether a metric follows its expected delayed response</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
@@ -641,7 +655,7 @@
         </text>
         <text class="meta" y="26">relationships/comparisons</text>
       </g>
-      <g class="node partial" data-capability="relationships/resolution/compare-finalized-subperiods" transform="translate(1198.1 733.3)">
+      <g class="node partial" data-capability="relationships/resolution/compare-finalized-subperiods" transform="translate(1138.7 709.7)">
         <title>Inspect related metrics before the business period closes</title>
         <circle class="node-halo" r="46" />
         <circle class="node-ring" r="38" />
diff --git a/.capabilities/dependency-viewer.html b/.capabilities/dependency-viewer.html
index 144ae27..1021aab 100644
--- a/.capabilities/dependency-viewer.html
+++ b/.capabilities/dependency-viewer.html
@@ -388,7 +388,7 @@ <h1>Capability dependency viewer</h1>
       <aside><div id="panel" class="panel"></div></aside>
     </div>
     <script>
-const graph = {"nodes":[{"id":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","label":"Understand how much histor...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","acceptance":["Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","Provisional residual scores can be reported but do not trigger point-anomaly episodes.","Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger."],"guidance":["Evidence strength is not a probability of an incident.","Calibration and expectations are separate concerns; each declared regime calibrates again."],"path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/detection/identify-residual-patterns","analysis/detection/report-point-anomaly-episodes","analysis/evidence/assess-reference-readiness","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","relationships/comparisons/compare-declared-metrics"],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"short-history","description":"Keep reference-only evidence for a short series.","command":"python -m pytest -q tests/test_analysis.py::test_short_series_returns_reference_evidence"},{"id":"provisional-evidence","description":"Report provisional scores without point triggers.","command":"python -m pytest -q tests/test_analysis.py::test_early_evidence_maturity_and_no_provisional_trigger"},{"id":"independent-calibration","description":"Check a known residual center, sample scale, and standardized score.","command":"python -m pytest -q tests/test_analysis.py::test_calibration_reference"},{"id":"constant-reference","description":"Handle zero residual variance without spurious episodes.","command":"python -m pytest -q tests/test_analysis.py::test_constant_and_schema"}],"manualChecks":["Inspect a sample's signal_maturity, calibration_samples, standardized_residual, and triggers together before describing a statistical departure."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","criteria":[{"criterion":"Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","status":"covered","evidence":["src/anomalyzer/recipes.py:205","src/anomalyzer/recipes.py:445","tests/test_analysis.py:141"],"notes":"The short-series test checks three reference-only samples, absent standardized scores, and no triggers."},{"criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"covered","evidence":["src/anomalyzer/recipes.py:210","tests/test_analysis.py:123"],"notes":"A provisional score of 2 exceeds the configured point threshold of 1 but produces no trigger or episode."},{"criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"covered","evidence":["src/anomalyzer/recipes.py:231","src/anomalyzer/recipes.py:307","tests/test_analysis.py:111"],"notes":"Calibration uses retained residual mean and sample standard deviation once per regime; the independent numerical example produces the expected standardized score."},{"criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"covered","evidence":["src/anomalyzer/contracts.py:104","src/anomalyzer/recipes.py:350","src/anomalyzer/recipes.py:388","tests/test_analysis.py:111"],"notes":"Sample fields retain the comparison and support; the point threshold is inspectable in the enclosing method parameters."}],"ignore_findings":[],"done":true,"input_fingerprint":"ccdc1b2e92f3e647d5a9e6127cc3c4b10bf0da8f1f78af3efb805daa9b1e4305","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Establish evidence support"},"impact":6,"gaps":0,"r":68,"x":170,"y":170},{"id":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","label":"Bring my metrics into an a...","area":"data","scopes":["data","data/inputs"],"scope":"data/inputs","scopeLabel":"data/inputs","status":"implemented","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","acceptance":["Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","Published schema 1.0 and 1.1 request and result schemas match the public Python contracts."],"guidance":["Null observations remain unknown unless a schema-1.1 additive-volume policy explicitly authorizes confirmed empty collection as zero. Limits apply to the request as a whole.","Describe current supported behavior; historical plans do not add accepted inputs."],"path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["analysis/evidence/calibrate-departures","analysis/expectations/model-seasonal-level-and-growth","data/quality/analyze-sparse-observations","execution/resilience/retain-evidence-under-failure-and-limits","interfaces/python/analyze-through-public-api","investigation/assessments/describe-early-changes"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_agentic.py","schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json"],"automatedChecks":[{"id":"input-forms","description":"Analyze a bare list and validate a canonical request and result.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings tests/test_analysis.py::test_constant_and_schema"},{"id":"explicit-scope","description":"Reject schema 1.1 datasets without entity scope.","command":"python -m pytest -q tests/test_agentic.py::test_schema_11_requires_explicit_entity_scope"},{"id":"invalid-inputs","description":"Reject unsupported settings and nonnumeric or nonfinite observations.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_config tests/test_analysis.py::test_invalid_values"},{"id":"published-schemas","description":"Compare all four checked-in schemas with their public contracts.","command":"python -m pytest -q tests/test_analysis.py::test_checked_in_schemas tests/test_agentic.py::test_v11_checked_in_schemas_exist"}],"manualChecks":["Review the input examples in README.md and confirm they describe the two supported schema versions and the four-dataset/four-relationship limit."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","criteria":[{"criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"covered","evidence":["src/anomalyzer/recipes.py:21","src/anomalyzer/contracts.py:96","tests/test_analysis.py:234"],"notes":"Bare-list settings are validated; schema 1.0 limits datasets to exactly one."},{"criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"covered","evidence":["src/anomalyzer/contracts.py:240","src/anomalyzer/contracts.py:335","tests/test_agentic.py:192"],"notes":"Entity scope is required, relationship kinds are discriminated, and references must name distinct known datasets."},{"criterion":"Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:7","src/anomalyzer/contracts.py:71","tests/test_analysis.py:526","tests/test_analysis.py:531"],"notes":"Assertions reject each listed category; missing null values remain a separate accepted representation."},{"criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"covered","evidence":["schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json","tests/test_analysis.py:547","tests/test_agentic.py:317"],"notes":"Tests and independent inspection compare each complete published schema with its contract-generated schema. Python cross-field validators supplement JSON Schema constraints."}],"ignore_findings":[],"done":true,"input_fingerprint":"3546cd8f436812ff11c9352a855926677b55cc6287e40f6a5041bd613173e5b4","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Supply a numeric series"},"impact":6,"gaps":0,"r":68,"x":353.33333333333337,"y":170},{"id":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","label":"Analyze local files from t...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","acceptance":["The methods command lists supported analysis recipes and residual detectors.","Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status."],"guidance":["CSV callers must select one entity and unit before analysis.","Exit 0 includes completed analysis with anomalies, insufficient history, or inapplicability; exit 2 means invalid input, and exit 1 means execution failure or partial execution."],"path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/python/install-versioned-library","interfaces/skill/guide-anomaly-investigation","interfaces/skill/visualize-analysis-on-request"],"implementationReferences":["src/anomalyzer/cli.py","src/anomalyzer/io.py","src/anomalyzer/__main__.py","tests/test_cli.py"],"automatedChecks":[{"id":"file-stdin","description":"Analyze equivalent file and stdin requests.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence"},{"id":"csv-precedence","description":"Validate selected CSV fields and apply flag-over-file-over-embedded precedence.","command":"python -m pytest -q tests/test_cli.py::test_csv_and_precedence"},{"id":"value-list","description":"Analyze a bare JSON array and expose inferred settings.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"},{"id":"output-and-exits","description":"Protect existing output and distinguish invalid options from failed execution.","command":"python -m pytest -q tests/test_cli.py::test_output_protection tests/test_cli.py::test_methods_and_invalid_options tests/test_cli.py::test_size_limit_and_execution_exit"}],"manualChecks":["Run an example in readable and JSON modes and confirm stdout contains results while diagnostics go to stderr."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","criteria":[{"criterion":"The methods command lists supported analysis recipes and residual detectors.","status":"covered","evidence":["src/anomalyzer/cli.py:426","src/anomalyzer/registry.py:15","tests/test_cli.py:27"],"notes":"Discovery tests check all three recipes and ten residual detectors, with structured stdout and no successful-run diagnostics on stderr."},{"criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:239","tests/test_cli.py:13","tests/test_mcp.py:163"],"notes":"The passing file/stdin regression compares request fingerprints and short-history semantics; cross-interface tests compare complete stable numerical output."},{"criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"covered","evidence":["src/anomalyzer/io.py:33","tests/test_cli.py:51"],"notes":"The passing CSV regression covers timestamped missing cells, positional data, and varying unmapped entity rejection."},{"criterion":"Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","status":"covered","evidence":["src/anomalyzer/cli.py:253","tests/test_cli.py:51","tests/test_cli.py:71"],"notes":"The passing file/flag/embedded precedence regression and bare-values regression establish the merge order and validated configuration."},{"criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"covered","evidence":["src/anomalyzer/io.py:69","src/anomalyzer/cli.py:424","tests/test_cli.py:41","tests/test_cli.py:108"],"notes":"Output protection and exit regressions passed, including explicit overwrite; successful manual examples have empty stderr and incomplete-usage exits 1."}],"ignore_findings":[],"done":true,"input_fingerprint":"86e701ccc0a1aefd407f11b4547784fef497fa1b49ddf6bb69db0d08f757dfdc","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Run from files or stdin"},"impact":4,"gaps":0,"r":58,"x":903.3333333333334,"y":170},{"id":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","label":"Account for ordinary seaso...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","acceptance":["A supplied season length selects earlier matching seasonal observations for the reference.","Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected."],"guidance":["Expectations support scoring; they are not a standalone forecasting API."],"path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/expectations/protect-reference-from-extremes","analysis/expectations/reset-declared-regimes","analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py"],"automatedChecks":[{"id":"seasonal-reference","description":"Check independent lagged-reference values.","command":"python -m pytest -q tests/test_analysis.py::test_forecast_independent_reference"},{"id":"growth-and-departures","description":"Keep clean linear and compound growth ordinary while identifying injected departures.","command":"python -m pytest -q tests/test_trend.py::test_trend_and_anomalies"},{"id":"frozen-fit","description":"Keep the training fit stable when later observations change.","command":"python -m pytest -q tests/test_trend.py::test_frozen_fit_and_legacy_mode"},{"id":"explicit-trends","description":"Accept supported trend choices and handle short, constant, and signed series.","command":"python -m pytest -q tests/test_trend.py::test_explicit_modes_and_validation tests/test_trend.py::test_short_history_constant_and_signed_linear"}],"manualChecks":["Compare examples/growth_up.json and examples/compound.json with their selected trend and evidence expectations."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","criteria":[{"criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"covered","evidence":["src/anomalyzer/baseline.py:8","src/anomalyzer/baseline.py:31","tests/test_analysis.py:21"],"notes":"Independent lag values and robust matching-phase assertions establish the seasonal reference calculation."},{"criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"covered","evidence":["src/anomalyzer/trend.py:120","src/anomalyzer/trend.py:144","tests/test_trend.py:11"],"notes":"Positive and negative linear/exponential cases select the expected model, match post-training expectations, and produce no ordinary-growth episodes."},{"criterion":"The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","status":"covered","evidence":["src/anomalyzer/recipes.py:162","tests/test_trend.py:36"],"notes":"Trend fitting occurs at the training boundary; the test asserts identical trend diagnostics and earlier evidence after later observations change."},{"criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:20","tests/test_trend.py:49","tests/test_trend.py:57"],"notes":"Tests exercise disabled, linear, exponential, and invalid quadratic modes, including disclosed short-history fallback."}],"ignore_findings":[],"done":true,"input_fingerprint":"8e9a0361b02b7d3f651366e5fa0675cccb3949e44de448d2ae0d43e6710bd408","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Account for growth and seasonality"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","label":"Embed local analysis in my...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","acceptance":["analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","analyze_series preserves supplied settings and event-time context for a single dataset.","analyze_relationships preserves the declared multi-dataset relationship semantics.","Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime."],"guidance":["All interfaces use local analysis; there is no hosted service or model API requirement."],"path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence","interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/__init__.py","src/anomalyzer/agent.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"request-parity","description":"Match canonical request results across Python, CLI, and full-result MCP compatibility calls.","command":"python -m pytest -q tests/test_mcp.py::test_canonical_request_matches_python_cli_and_mcp tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"series-context","description":"Preserve a single-series event-time cutoff and context across interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp tests/test_mcp.py::test_series_wrapper_preserves_omitted_and_explicit_season_settings"},{"id":"relationship-parity","description":"Match declared relationship evidence across the public interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_relationship_semantics_match_python_cli_and_mcp"}],"manualChecks":["Review the Python API examples in README.md and confirm callers can inspect the returned validated models and JSON."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","criteria":[{"criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"covered","evidence":["src/anomalyzer/__init__.py:3","src/anomalyzer/recipes.py:21","tests/test_analysis.py:234"],"notes":"Bare values and separate settings validate before dispatch; canonical schema 1.0 and 1.1 requests use their respective contracts."},{"criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"covered","evidence":["src/anomalyzer/agent.py:11","tests/test_mcp.py:191","tests/test_mcp.py:291"],"notes":"Field-set handling preserves omitted versus explicit settings and context in the single-series wrapper, verified across dictionary and Settings inputs."},{"criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"covered","evidence":["src/anomalyzer/agent.py:27","tests/test_mcp.py:217"],"notes":"Declared relationships validate and use the shared engine without changing units, lineage, or primitive semantics."},{"criterion":"Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime.","status":"covered","evidence":["tests/test_mcp.py:163","tests/test_mcp.py:191","tests/test_mcp.py:217","src/anomalyzer/__main__.py:1"],"notes":"Passing real CLI subprocess, Python, and full-result MCP parity tests compare stable numerical results apart from identity and runtime."}],"ignore_findings":[],"done":true,"input_fingerprint":"214bd10bb3aa5fad0f74938fe2ef09853a593cdb5b93cba49ec851a628d8ab65","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","name":"Python integration","label":"Python integration / Embed analysis in my application / Call the public API"},"impact":3,"gaps":0,"r":53,"x":903.3333333333334,"y":170},{"id":"investigation/cases/group-related-evidence","title":"Investigate related findings together","label":"Investigate related findin...","area":"investigation","scopes":["investigation","investigation/cases"],"scope":"investigation/cases","scopeLabel":"investigation/cases","status":"implemented","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","acceptance":["Evidence for different declared entities remains in separate cases even at the same event time.","Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","A derived relationship adds context without counting as an additional independent source for severity.","Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection."],"guidance":["Warning and critical are deterministic numerical grouping rules, not validated incident impact.","Diagnostic trend, oscillation, mixture, and variation findings remain distinct from adaptive location-shift cases."],"path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","dependencies":["investigation/assessments/describe-early-changes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/mcp/investigate-findings-and-evidence","investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/cases.py","src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py"],"automatedChecks":[{"id":"entity-scope","description":"Separate cases for different entities at the same event time.","command":"python -m pytest -q tests/test_agentic.py::test_cases_are_scoped_by_entity_at_the_same_event_time"},{"id":"source-severity","description":"Escalate numerical severity when two source datasets meet criteria.","command":"python -m pytest -q tests/test_agentic.py::test_two_source_datasets_escalate_case_severity"},{"id":"correlated-context","description":"Keep relationship evidence from double-counting source severity.","command":"python -m pytest -q tests/test_agentic.py::test_relationship_evidence_explains_but_does_not_double_count_severity"},{"id":"first-detection","description":"Group an adaptive location shift before point triggers without using a later pattern peak.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_location_shift_creates_one_early_case_without_point_trigger tests/test_agentic.py::test_location_case_severity_does_not_use_future_pattern_peak"}],"manualChecks":["Inspect a case's contributing targets, explanation, evidence references, and investigation questions; distinguish it from a point-anomaly episode."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","criteria":[{"criterion":"Evidence for different declared entities remains in separate cases even at the same event time.","status":"covered","evidence":["src/anomalyzer/cases.py:42","tests/test_agentic.py:69"],"notes":"Entity participates in grouping and stable identity; the test requires separate cases and distinct IDs at the same time."},{"criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"covered","evidence":["src/anomalyzer/cases.py:72","tests/test_agentic.py:85"],"notes":"Two qualifying source datasets produce critical numerical severity and named source explanations."},{"criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"covered","evidence":["src/anomalyzer/cases.py:72","src/anomalyzer/cases.py:116","tests/test_agentic.py:100"],"notes":"The independent-source count includes dataset criteria only. Derived evidence remains explanatory and can still meet the separately declared magnitude rule."},{"criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"covered","evidence":["src/anomalyzer/early.py:193","src/anomalyzer/early.py:237","src/anomalyzer/orchestrator.py:115","tests/test_agentic.py:122","tests/test_agentic.py:146"],"notes":"Tests require one moderate-shift case without point triggers and restrict assessment references/strength to samples at or before first detection."}],"ignore_findings":[],"done":true,"input_fingerprint":"6830af8a2f064123c12adb062f4793e446b97b3b805fd73a934ea84cd0d5d9cf","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Investigate related evidence"},"impact":3,"gaps":0,"r":53,"x":1086.6666666666667,"y":170},{"id":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","label":"Spot sustained shifts and ...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","acceptance":["Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","Trend, oscillation, and mixture findings retain their specific pattern meanings.","Moving-range findings identify variation changes separately from location changes."],"guidance":["Pattern findings may overlap and are not independent confirmations.","A residual location shift does not establish a change in the full distribution."],"path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations","investigation/cases/group-related-evidence"],"implementationReferences":["src/anomalyzer/evidence.py","src/anomalyzer/recipes.py","tests/test_analysis.py"],"automatedChecks":[{"id":"location-shift","description":"Find moderate increases and decreases without point anomalies.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_patterns_describe_location_shift_without_point_anomaly"},{"id":"accumulating-change","description":"Detect a known CUSUM accumulation at the expected position.","command":"python -m pytest -q tests/test_analysis.py::test_cusum_detects_accumulated_moderate_location_shift"},{"id":"diagnostic-patterns","description":"Distinguish residual trend, oscillation, and mixture rules.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_diagnostics_have_specific_semantics"},{"id":"variation","description":"Describe a moving-range increase as variation rather than location.","command":"python -m pytest -q tests/test_analysis.py::test_moving_range_detects_short_term_variation_increase"}],"manualChecks":["Compare examples/location_shift_up.json with its point episodes and pattern findings, allowing for overlapping residual rules."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","criteria":[{"criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"covered","evidence":["src/anomalyzer/evidence.py:118","src/anomalyzer/evidence.py:189","tests/test_analysis.py:293"],"notes":"Tests assert no point triggers or episodes and exact upward/downward detection positions for Nelson rules and CUSUM."},{"criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"covered","evidence":["src/anomalyzer/evidence.py:211","src/anomalyzer/evidence.py:230","tests/test_analysis.py:350"],"notes":"The known accumulation detects at index 16 and reports peak CUSUM approximately 6, threshold 5, and increase direction."},{"criterion":"Trend, oscillation, and mixture findings retain their specific pattern meanings.","status":"covered","evidence":["src/anomalyzer/evidence.py:70","src/anomalyzer/evidence.py:122","tests/test_analysis.py:340"],"notes":"Separate predicates and descriptions preserve the three meanings and explicitly limit location/distribution claims."},{"criterion":"Moving-range findings identify variation changes separately from location changes.","status":"covered","evidence":["src/anomalyzer/evidence.py:234","tests/test_analysis.py:360"],"notes":"Adjacent standardized-residual differences produce variation_shift findings; the test checks magnitude, detection index, and interpretation."}],"ignore_findings":[],"done":true,"input_fingerprint":"63716eb67db6b73964b79897d8c81f2891df2586ff213c95ea81c5699b8abbea","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Find sustained residual changes"},"impact":2,"gaps":0,"r":48,"x":170,"y":170},{"id":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","label":"Analyze observations in th...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","acceptance":["Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration."],"guidance":["No interpolation or implicit filling, resampling, or unit conversion occurs. Schema-1.1 source policies may explicitly authorize confirmed empty additive volume as zero.","Duplicate null values propagate through aggregation."],"path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["data/quality/exclude-incomplete-periods","relationships/quality/require-compatible-sources"],"implementationReferences":["src/anomalyzer/sparse.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/contracts.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"coordinate-contract","description":"Reject partial timestamp metadata and positional as_of cutoffs.","command":"python -m pytest -q tests/test_analysis.py::test_timestamp_metadata_is_all_or_nothing"},{"id":"local-time","description":"Normalize valid local times and reject ambiguous or nonexistent times.","command":"python -m pytest -q tests/test_analysis.py::test_local_time"},{"id":"duplicates-cutoff","description":"Require an explicit duplicate policy and retain counts after the cutoff.","command":"python -m pytest -q tests/test_analysis.py::test_duplicate_sort_and_cutoff"},{"id":"cadence-context","description":"Check fixed weekly cadence and preserve cutoff semantics across interfaces.","command":"python -m pytest -q tests/test_analysis.py::test_weekly_cadence tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp"}],"manualChecks":["Inspect a timestamped result's data_quality transformations, UTC window, and resolved context to confirm preparation is explainable."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","criteria":[{"criterion":"Timestamped datasets provide both timestamps and frequency; positional datasets provide neither.","status":"covered","evidence":["src/anomalyzer/contracts.py:77","tests/test_analysis.py:418"],"notes":"Partial coordinate metadata and positional as_of are rejected."},{"criterion":"Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","status":"covered","evidence":["src/anomalyzer/validation.py:8","tests/test_analysis.py:500"],"notes":"UTC round trips detect ambiguous/nonexistent local times; valid local time normalizes to the expected UTC hour."},{"criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"covered","evidence":["src/anomalyzer/validation.py:87","src/anomalyzer/validation.py:103","tests/test_analysis.py:478"],"notes":"Explicit mean/sum handling retains duplicate counts and transformation descriptions. Nulls propagate rather than becoming zero."},{"criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"covered","evidence":["src/anomalyzer/validation.py:105","src/anomalyzer/validation.py:111","tests/test_analysis.py:491","tests/test_mcp.py:191"],"notes":"Filtering is inclusive through as_of; cadence compares UTC elapsed timedeltas, including fixed seven-day weeks."}],"ignore_findings":[],"done":true,"input_fingerprint":"791748764cdc0dfe81d987a908179930d858670606e51b64407172295776b261","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Prepare the comparison"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","label":"Let an agent inspect the e...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","acceptance":["The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","get_evidence exposes the selected finding's numerical support and relationship source lineage.","Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated."],"guidance":["Run results are retained only in the bounded server-session cache; missing or evicted run IDs require reanalysis.","Cases, episodes, and residual patterns may overlap."],"path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":["interfaces/mcp/connect-read-only-local-server","interfaces/skill/guide-anomaly-investigation"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py"],"automatedChecks":[{"id":"case-drilldown","description":"Analyze, page findings, and inspect exact conversion case values and lineage.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case tests/test_mcp.py::test_schema_11_episode_is_paged_and_has_relationship_evidence"},{"id":"unavailable-finding","description":"Expose a quality finding when latest relationship evidence is unavailable.","command":"python -m pytest -q tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"},{"id":"early-candidate","description":"Inspect descriptive candidate evidence without inventing a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case tests/test_mcp.py::test_sparse_insufficient_history_keeps_descriptive_candidate_visible"},{"id":"sample-pagination","description":"Page supporting samples for a residual pattern.","command":"python -m pytest -q tests/test_mcp.py::test_get_evidence_pages_pattern_samples"}],"manualChecks":["Follow analyze, list_findings, and get_evidence for the conversion example and check that the headline is supported by the retrieved values."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","criteria":[{"criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"covered","evidence":["src/anomalyzer/mcp.py:170","src/anomalyzer/mcp_views.py:247","tests/test_mcp.py:54"],"notes":"The conversion manual call returns a relationship headline, preview, counts, readiness, and session-scoped run ID."},{"criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:132","src/anomalyzer/mcp_views.py:260","tests/test_mcp.py:313","tests/test_mcp.py:108"],"notes":"Cases, early candidates, episodes, patterns, and quality findings feed a bounded page; relationship episode regression verifies schema 1.1 pagination."},{"criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:273","src/anomalyzer/mcp_views.py:294","tests/test_mcp.py:54","tests/test_mcp.py:313"],"notes":"Manual drilldown returns observed 8/130 versus expected 0.1 and source indexes orders=14 and qualified_visits=14."},{"criterion":"Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:118","src/anomalyzer/mcp_views.py:169","src/anomalyzer/mcp_views.py:337","tests/test_mcp.py:87","tests/test_mcp.py:137","tests/test_mcp.py:341"],"notes":"Passing tests retain short-history and inapplicable-method candidates plus unavailable latest evidence, while pattern samples are paginated."}],"ignore_findings":[],"done":true,"input_fingerprint":"a09b0003780bcd2f9ccf8e3d8ff58b3147c118760b984757d7027b5810fd9bef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Drill into findings and evidence"},"impact":2,"gaps":0,"r":48,"x":927.3333333333334,"y":288},{"id":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","label":"Spot early changes with a ...","area":"investigation","scopes":["investigation","investigation/assessments"],"scope":"investigation/assessments","scopeLabel":"investigation/assessments","status":"implemented","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","acceptance":["The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","Missing values break the comparison history, and zero previous values do not create relative-change evidence.","Early candidates remain non-triggering and ineligible for action even under a permissive policy."],"guidance":["An early candidate does not establish an anomaly, stable baseline, cause, or incident."],"path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["investigation/cases/group-related-evidence","investigation/rules/check-declared-numerical-rules"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_mcp.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"prefix-invariant-facts","description":"Describe short history causally without creating cases.","command":"python -m pytest -q tests/test_agentic.py::test_short_history_assessments_are_useful_and_prefix_invariant"},{"id":"comparison-boundaries","description":"Reset comparisons at missing observations and avoid dividing by zero.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"candidate-eligibility","description":"Keep candidates ineligible for action and notifications under permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"},{"id":"agent-visibility","description":"Expose a short-history candidate and its claim limits without a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case"}],"manualChecks":["Inspect a candidate's comparison population, maturity, established claim, and not_established text before describing it."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","criteria":[{"criterion":"The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","status":"covered","evidence":["src/anomalyzer/early.py:73","src/anomalyzer/early.py:96","tests/test_agentic.py:35"],"notes":"Assertions establish observation/change/change/departure_candidate for [10,12,11,20]."},{"criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"covered","evidence":["src/anomalyzer/early.py:50","src/anomalyzer/early.py:141","tests/test_agentic.py:35"],"notes":"Each assessment precedes the prior-history update; tests compare complete serialized assessments for every prefix."},{"criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"covered","evidence":["src/anomalyzer/early.py:53","src/anomalyzer/early.py:58","tests/test_capability_behaviors.py:87"],"notes":"The zero comparison has no relative change; after a missing value, comparison samples reset to zero."},{"criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"covered","evidence":["src/anomalyzer/early.py:105","src/anomalyzer/cases.py:35","tests/test_capability_behaviors.py:134","tests/test_mcp.py:341"],"notes":"Candidates have criterion_met=false and remain action/notification-ineligible after policy replay, without creating cases."}],"ignore_findings":[],"done":true,"input_fingerprint":"debdeeeaea559950f53461a4298cb9c2e3a1681f83a43e8a981ed9b51968a4d0","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect early evidence"},"impact":2,"gaps":0,"r":48,"x":1110.6666666666667,"y":288},{"id":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","label":"Compare the same entity an...","area":"relationships","scopes":["relationships","relationships/quality"],"scope":"relationships/quality","scopeLabel":"relationships/quality","status":"implemented","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","acceptance":["Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","Difference relationships require identical declared source units and matching output units.","An invalid relationship is reported as failed while valid sibling dataset results remain available."],"guidance":["A failed or unavailable relationship does not establish that its metric is normal.","No implicit timestamp matching, interpolation, unit conversion, or zero-filling occurs."],"path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"implementationReferences":["src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"exact-alignment","description":"Reject mismatched timestamps, cadence, entity, duplicates, and positional relationship sources.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relationship_requires_exact_source_alignment"},{"id":"compatible-units","description":"Report a difference unit mismatch without suppressing sibling results.","command":"python -m pytest -q tests/test_agentic.py::test_alignment_unit_and_partial_failure_are_specific"},{"id":"unambiguous-declaration","description":"Reject fields that do not belong to the selected relationship kind.","command":"python -m pytest -q tests/test_agentic.py::test_contract_rejects_ambiguous_relationship_fields"},{"id":"healthy-siblings","description":"Retain a healthy dataset's evidence when another dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Read both target-level statuses and errors before interpreting a partial run; check source dataset errors when a relationship source failed."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","criteria":[{"criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"covered","evidence":["src/anomalyzer/alignment.py:42","tests/test_capability_behaviors.py:30"],"notes":"Parameterized assertions require specific errors for timestamp, cadence, and entity mismatches."},{"criterion":"Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","status":"covered","evidence":["src/anomalyzer/alignment.py:23","src/anomalyzer/alignment.py:44","tests/test_capability_behaviors.py:30"],"notes":"Positional sources fail exact alignment; even an explicit duplicate-mean setting is rejected for relationship source lineage."},{"criterion":"Difference relationships require identical declared source units and matching output units.","status":"covered","evidence":["src/anomalyzer/relationships.py:41","tests/test_agentic.py:242"],"notes":"Source and output unit checks are explicit. Tests cover mismatched source units and successful matching units."},{"criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:171","tests/test_capability_behaviors.py:30","tests/test_capability_behaviors.py:159"],"notes":"Relationship errors remain target-specific, healthy dataset results survive, and the overall mixed run is partial."}],"ignore_findings":[],"done":true,"input_fingerprint":"06321ab5806a8276cc3954df5d6f4dd7f059f67600038854569787ff51b420d6","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Align comparable sources"},"impact":2,"gaps":0,"r":48,"x":1294,"y":288},{"id":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","label":"Compare against a new base...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","acceptance":["Integer positions and exact observed dates or timestamps can identify reset boundaries.","Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","Repeated, out-of-range, or unmatched reset points are rejected.","CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries."],"guidance":["Reset points are explicit caller decisions. Known-event notes and review-state fields do not rebuild regimes.","Multi-resolution analysis does not support reset_points."],"path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations"],"implementationReferences":["src/anomalyzer/validation.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py"],"automatedChecks":[{"id":"independent-regimes","description":"Retrain separate regimes without cross-boundary evidence or patterns.","command":"python -m pytest -q tests/test_analysis.py::test_manual_resets_retrain_without_crossing_boundaries"},{"id":"invalid-boundaries","description":"Reject invalid or duplicate resolved reset positions.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_manual_resets"},{"id":"cli-boundaries","description":"Resolve position and date reset flags through the CLI.","command":"python -m pytest -q tests/test_cli.py::test_manual_reset_cli"}],"manualChecks":["Inspect the resolved reset positions and segment diagnostics for examples/steps_reset.json."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","criteria":[{"criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"covered","evidence":["src/anomalyzer/validation.py:129","tests/test_analysis.py:381"],"notes":"Positions and observed dates are exercised end to end; timestamp normalization and exact matching are implemented directly."},{"criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"covered","evidence":["src/anomalyzer/recipes.py:119","src/anomalyzer/recipes.py:143","tests/test_analysis.py:381"],"notes":"Each segment initializes separate state and passes its own start to forecasts. Tests verify three calibrated regimes and no cross-boundary evidence or findings."},{"criterion":"Repeated, out-of-range, or unmatched reset points are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/validation.py:149","tests/test_analysis.py:413"],"notes":"Validation rejects literal duplicates and different declarations resolving to the same position, invalid positions, and unmatched dates/timestamps."},{"criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"covered","evidence":["tests/test_cli.py:82","src/anomalyzer/recipes.py:53"],"notes":"CLI assertions establish positions [50, 100] and removal of the unmarked-regime episodes."}],"ignore_findings":[],"done":true,"input_fingerprint":"067063f972c5b70d990d6cb0125892731a97a7f482698d7b331a1a30229ac223","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Restart after a reviewed change"},"impact":1,"gaps":0,"r":43,"x":194,"y":288},{"id":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","label":"Assess sparse observations...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","acceptance":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"guidance":["Default missingness is unknown. A modeled expectation is never a measurement.","Caller business severity, incident state and notifications remain outside the engine."],"path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","src/anomalyzer/orchestrator.py","src/anomalyzer/relationships.py","src/anomalyzer/early.py","tests/test_sparse.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"source-semantics","description":"Preserve unknown periods and require explicit empty-volume authorization.","command":"python -m pytest -q tests/test_sparse.py -k \"zero or omitted or grid or duplicate or expansion\""},{"id":"causal-boundaries","description":"Retain real phase, event times, resets and as-of revision eligibility.","command":"python -m pytest -q tests/test_sparse.py -k \"phase or gap or reset or event or revision or arrival\""},{"id":"public-interfaces","description":"Preserve useful rules, relationships, views, cases and API parity.","command":"python -m pytest -q tests/test_sparse.py -k \"rule or relationship or daily or parity\""},{"id":"progressive-examples","description":"Reproduce the full numbered sequence and sparse worked examples.","command":"python examples/run_demo.py"}],"manualChecks":["Inspect progression figures 14–22 for true gaps, distinguished policy zeros, original event time, unavailable checks and caller-owned business severity."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/analyze-sparse-observations.","criteria":[{"criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","status":"partial","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:71","src/anomalyzer/contracts.py:85","src/anomalyzer/contracts.py:86","src/anomalyzer/sparse.py:3","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:42","src/anomalyzer/sparse.py:48","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse_views.py:74","src/anomalyzer/sparse_views.py:81","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:97","src/anomalyzer/sparse_views.py:119","src/anomalyzer/orchestrator.py:63","src/anomalyzer/orchestrator.py:127","src/anomalyzer/orchestrator.py:139","src/anomalyzer/orchestrator.py:191","src/anomalyzer/relationships.py:36","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:98","src/anomalyzer/relationships.py:131","src/anomalyzer/relationships.py:141","src/anomalyzer/early.py:58","src/anomalyzer/early.py:137","tests/test_sparse.py:44","tests/test_sparse.py:45","tests/test_sparse.py:50","tests/test_sparse.py:51","tests/test_sparse.py:57","docs/sparse-data.md:4","docs/sparse-data.md:29","docs/sparse-data.md:37","docs/sparse-data.md:39","docs/sparse-data.md:40","examples/sparse_examples.py:22","examples/sparse_examples.py:23","examples/sparse_examples.py:25","examples/sparse_examples.py:26","examples/sparse_examples.py:28","examples/render_readme_progression.py:135","examples/render_readme_progression.py:139","examples/render_readme_progression.py:140","examples/render_readme_progression.py:166","examples/render_readme_progression.py:206"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","status":"partial","evidence":["src/anomalyzer/contracts.py:27","src/anomalyzer/contracts.py:49","src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:56","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:181","src/anomalyzer/sparse.py:187","src/anomalyzer/sparse.py:189","src/anomalyzer/sparse_views.py:15","src/anomalyzer/sparse_views.py:98","src/anomalyzer/sparse_views.py:107","src/anomalyzer/sparse_views.py:108","src/anomalyzer/sparse_views.py:112","src/anomalyzer/orchestrator.py:22","src/anomalyzer/orchestrator.py:23","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:129","src/anomalyzer/orchestrator.py:141","src/anomalyzer/relationships.py:185","src/anomalyzer/relationships.py:189","src/anomalyzer/relationships.py:191","src/anomalyzer/early.py:46","src/anomalyzer/early.py:48","src/anomalyzer/early.py:51","src/anomalyzer/early.py:59","src/anomalyzer/early.py:61","tests/test_sparse.py:20","tests/test_sparse.py:108","tests/test_sparse.py:111","tests/test_sparse.py:113","tests/test_sparse.py:119","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:42","docs/sparse-data.md:106","docs/sparse-data.md:108","examples/sparse_examples.py:16","examples/sparse_examples.py:33","examples/sparse_examples.py:50","examples/sparse_examples.py:51","examples/sparse_examples.py:91","examples/render_readme_progression.py:241","examples/render_readme_progression.py:312","examples/render_readme_progression.py:317","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:242","src/anomalyzer/contracts.py:245","src/anomalyzer/contracts.py:258","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:57","src/anomalyzer/sparse.py:63","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:127","src/anomalyzer/sparse_views.py:155","src/anomalyzer/sparse_views.py:157","src/anomalyzer/orchestrator.py:37","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:178","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:40","src/anomalyzer/relationships.py:57","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:96","src/anomalyzer/relationships.py:97","src/anomalyzer/early.py:140","src/anomalyzer/early.py:147","tests/test_sparse.py:59","tests/test_sparse.py:67","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:128","docs/sparse-data.md:26","docs/sparse-data.md:96","docs/sparse-data.md:101","docs/sparse-data.md:132","docs/sparse-data.md:142","examples/sparse_examples.py:26","examples/sparse_examples.py:36","examples/sparse_examples.py:38","examples/sparse_examples.py:57","examples/sparse_examples.py:81","examples/render_readme_progression.py:654","examples/render_readme_progression.py:697","examples/render_readme_progression.py:884","examples/render_readme_progression.py:896","examples/render_readme_progression.py:917"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","status":"partial","evidence":["src/anomalyzer/contracts.py:13","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:297","src/anomalyzer/contracts.py:298","src/anomalyzer/contracts.py:437","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse.py:66","src/anomalyzer/sparse.py:83","src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:28","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:94","src/anomalyzer/orchestrator.py:106","src/anomalyzer/orchestrator.py:107","src/anomalyzer/orchestrator.py:133","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:160","src/anomalyzer/relationships.py:162","src/anomalyzer/relationships.py:166","tests/test_sparse.py:59","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:123","tests/test_sparse.py:180","docs/sparse-data.md:26","docs/sparse-data.md:41","docs/sparse-data.md:69","docs/sparse-data.md:78","docs/sparse-data.md:96","examples/sparse_examples.py:59","examples/sparse_examples.py:95","examples/render_readme_progression.py:686","examples/render_readme_progression.py:699","examples/render_readme_progression.py:710","examples/render_readme_progression.py:767","examples/render_readme_progression.py:884"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","partial: Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","partial: Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","partial: As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Assess sparse source data"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","label":"Avoid mistaking unfinished...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","acceptance":["Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","The observed values and timestamps of unfinished periods remain available in data quality output.","Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","A relationship with an unavailable latest source reports that unavailability rather than normal behavior."],"guidance":["Completeness is declared by the caller; the engine does not infer collection completeness."],"path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/alignment.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"unfinished-values","description":"Retain an unfinished value without scoring it.","command":"python -m pytest -q tests/test_analysis.py::test_explicit_incomplete_period_is_reported_and_not_scored"},{"id":"trailing-only","description":"Reject incomplete observations followed by completed observations.","command":"python -m pytest -q tests/test_analysis.py::test_incomplete_periods_must_be_trailing"},{"id":"source-and-agent-views","description":"Exclude a partial source sample from both analysis views and expose unavailable relationship evidence.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"}],"manualChecks":["Inspect examples/agentic/incomplete-usage.json and confirm the latest headline explains missing completeness and does not imply normality."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicitly incomplete trailing observations are reported as data quality evidence and excluded from completed-series analysis.","criteria":[{"criterion":"Incomplete observations form a trailing suffix after chronological preparation.","status":"covered","evidence":["src/anomalyzer/validation.py:44","tests/test_analysis.py:162"],"notes":"Suffix validation runs after timestamp preparation and rejects a completed observation following an incomplete one."},{"criterion":"The observed values and timestamps of unfinished periods remain available in data quality output.","status":"covered","evidence":["src/anomalyzer/validation.py:50","src/anomalyzer/validation.py:114","tests/test_analysis.py:150"],"notes":"Quality records retain original source index, normalized timestamp, observed value, and incomplete status."},{"criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"covered","evidence":["src/anomalyzer/validation.py:53","src/anomalyzer/recipes.py:564","tests/test_analysis.py:150","tests/test_analysis.py:200"],"notes":"Only finalized prepared values enter either scoring path; assertions verify the last scored sample and both multi-resolution counts."},{"criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"covered","evidence":["src/anomalyzer/alignment.py:35","src/anomalyzer/orchestrator.py:150","tests/test_mcp.py:87"],"notes":"Incomplete aligned sources become unavailable values, and latest relationship state/status plus MCP headline explain the lack of evidence."}],"ignore_findings":[],"done":true,"input_fingerprint":"d82ee816fe89a25232e2589587d31bcadfa81be4188a0e1531f4fbfdf0ebcb5e","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Keep unfinished data visible"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","label":"Give an agent read-only ac...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","acceptance":["Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","Analysis tool calls return structured content together with compact text.","Invalid tool input and unknown session run IDs return structured errors rather than fabricated results."],"guidance":["Compatibility tool names remain callable but are not advertised.","Legacy clients receive isError tool results; modern clients receive JSON-RPC invalid-input errors.","The server provides analysis and evidence inspection; the caller owns monitoring and operational workflows."],"path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","dependencies":["interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":["interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py","pyproject.toml"],"automatedChecks":[{"id":"tool-discovery","description":"Advertise the three read-only tools and their schema contracts.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_lists_read_only_tools"},{"id":"connection-protocols","description":"Support legacy stdio initialization and modern stateless discovery.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_stdio_legacy_handshake_and_tool_listing tests/test_mcp.py::test_mcp_modern_discovery_and_stateless_tool_listing"},{"id":"structured-content","description":"Return a compact analysis response with structured content and run ID.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_tool_call_returns_structured_content"},{"id":"invalid-calls","description":"Reject malformed input and unknown session run IDs in both protocols.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_invalid_tool_input_is_a_tool_error tests/test_mcp.py::test_paged_tool_inputs_enforce_schema_in_both_protocols"}],"manualChecks":["Review discovery's interpretation guidance and confirm the client can start anomalyzer-mcp in the installed Python environment."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","criteria":[{"criterion":"Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","status":"covered","evidence":["src/anomalyzer/mcp.py:63","tests/test_mcp.py:27"],"notes":"Discovery exposes three read-only tools with the same strict argument contracts used at dispatch and advertised output schemas."},{"criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"covered","evidence":["src/anomalyzer/mcp.py:228","tests/test_mcp.py:248","tests/test_mcp.py:268"],"notes":"Legacy initialization and modern stateless discovery tests establish protocol metadata and evidence interpretation guidance."},{"criterion":"Analysis tool calls return structured content together with compact text.","status":"covered","evidence":["src/anomalyzer/mcp.py:162","tests/test_mcp.py:39"],"notes":"The manual conversion call returns a compact headline and structured run ID, counts, readiness, and preview."},{"criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"covered","evidence":["src/anomalyzer/mcp.py:156","src/anomalyzer/mcp.py:257","tests/test_mcp.py:235","tests/test_mcp.py:367"],"notes":"Tests reject invalid cursor/limit types, unknown fields, numeric run IDs, and missing session runs in both protocols without fabricated results."}],"ignore_findings":[],"done":true,"input_fingerprint":"7bd92d238f4f9545372c6c35e78aefea60c585bea0a4a8bb99096ab6cef27f96","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Connect a local server"},"impact":1,"gaps":0,"r":43,"x":903.3333333333334,"y":406},{"id":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","label":"Find values that break my ...","area":"investigation","scopes":["investigation","investigation/rules"],"scope":"investigation/rules","scopeLabel":"investigation/rules","status":"implemented","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","acceptance":["Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","Absolute and relative changes must exceed their declared maximum to violate a rule.","Relative-change checks require a nonzero previous value and do not bridge missing observations.","Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them."],"guidance":["An exact rule crossing does not establish cause, an incident, or business impact."],"path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","dependencies":["investigation/assessments/describe-early-changes"],"suggestedDependencies":[],"dependents":["investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"rule-boundaries","description":"Check inclusive range boundaries and strict absolute/relative change thresholds.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_explicit_rules_respect_boundaries"},{"id":"relative-availability","description":"Avoid relative violations across zero or missing values.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"explicit-not-actionable","description":"Create evidence and a case for an exact rule crossing without enabling action by default.","command":"python -m pytest -q tests/test_agentic.py::test_explicit_rule_is_exact_but_not_actionable_without_policy"},{"id":"readable-rule","description":"Identify the exact unit-cost range rule in CLI output.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the unit-cost example's crossed rule and confirm its text distinguishes the declared bound from statistical detection."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","criteria":[{"criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"covered","evidence":["src/anomalyzer/early.py:22","src/anomalyzer/contracts.py:209","tests/test_capability_behaviors.py:74"],"notes":"Boundary values 10 and 20 are accepted; 9 and 21 violate the declared range."},{"criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"covered","evidence":["src/anomalyzer/early.py:25","tests/test_capability_behaviors.py:74"],"notes":"Parameterized tests distinguish equality to the threshold from strict exceedance for both change rules."},{"criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"covered","evidence":["src/anomalyzer/early.py:29","src/anomalyzer/early.py:53","tests/test_capability_behaviors.py:87"],"notes":"Zero/missing predecessors suppress relative comparisons, and missing values reset comparison history."},{"criterion":"Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them.","status":"covered","evidence":["src/anomalyzer/early.py:34","src/anomalyzer/early.py:121","tests/test_agentic.py:49","tests/test_cli.py:134"],"notes":"Violation records retain rule ID/kind/bounds; default eligibility is false and replay requires explicit permission."}],"ignore_findings":[],"done":true,"input_fingerprint":"60d24e84faee9c1e2ae8f9337d17334b37c89305093404da2e84c51526111d0b","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Apply my stated limits"},"impact":1,"gaps":0,"r":43,"x":1086.6666666666667,"y":406},{"id":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","label":"Find changes in conversion...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","acceptance":["Ratios retain the computed value, formula, and original source indexes for each sample.","Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","Derived series can be analyzed with the frozen or adaptive single-resolution recipe."],"guidance":["Relationships are caller-supplied declarations, not discovered or causal relationships.","A difference requires identical declared units; a normalized residual requires a positive supplied scale.","A joint condition is a declared 0/1 comparison, not statistical proof."],"path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","dependencies":["relationships/quality/require-compatible-sources","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_examples.py"],"automatedChecks":[{"id":"ratio-lineage","description":"Explain a changing rate with the exact source indexes and formula.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"other-comparisons","description":"Check declared differences, normalized residuals, and combined conditions.","command":"python -m pytest -q tests/test_agentic.py::test_remaining_relationship_primitives"},{"id":"unavailable-values","description":"Keep zero-denominator and missing-source observations unavailable.","command":"python -m pytest -q tests/test_agentic.py::test_missing_and_zero_denominator_are_unavailable_not_zero"},{"id":"adaptive-relationship","description":"Use the adaptive recipe for a declared relationship.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_recipe_runs_for_declared_relationships"}],"manualChecks":["Inspect examples/agentic/conversion.json and distinguish its relationship departure from the individual source metrics."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","criteria":[{"criterion":"Ratios retain the computed value, formula, and original source indexes for each sample.","status":"covered","evidence":["src/anomalyzer/relationships.py:84","src/anomalyzer/relationships.py:152","tests/test_agentic.py:211"],"notes":"Assertions check exact ratio values, formula, and both original source indexes."},{"criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"covered","evidence":["src/anomalyzer/relationships.py:97","src/anomalyzer/relationships.py:134","tests/test_agentic.py:285"],"notes":"Parameterized assertions independently specify difference, normalized residual, all-condition, and any-condition outputs."},{"criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"covered","evidence":["src/anomalyzer/relationships.py:89","src/anomalyzer/relationships.py:142","tests/test_agentic.py:226"],"notes":"Lineage retains null values with missing_source_value/zero_denominator reasons. Explicit zero_denominator=error intentionally fails instead, as documented."},{"criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:87","src/anomalyzer/orchestrator.py:135","tests/test_agentic.py:171","tests/test_examples.py:17"],"notes":"Both supported recipes dispatch derived datasets through the shared single-resolution engine."}],"ignore_findings":[],"done":true,"input_fingerprint":"7fa8f7ffdc0d852ad961b3e1753761d1bd3b00b87b8c001a957ec19e26d068b9","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Compare ratios and metric gaps"},"impact":1,"gaps":0,"r":43,"x":1270,"y":406},{"id":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","label":"Find unusual spikes and dr...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","acceptance":["A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","Separated point departures remain separate episodes, and positional episode intervals use integer coordinates."],"guidance":["Detection is a numerical departure from the selected expectation, not a claim about cause or business impact."],"path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/evidence.py","tests/test_analysis.py"],"automatedChecks":[{"id":"point-evidence","description":"Detect an injected spike and retain deterministic earlier numerical evidence.","command":"python -m pytest -q tests/test_analysis.py::test_spike_and_no_lookahead"},{"id":"adjacent-run","description":"Describe adjacent point triggers and their contributing positions.","command":"python -m pytest -q tests/test_analysis.py::test_anomaly_stream_detects_consecutive_run"},{"id":"isolated-events","description":"Keep separated anomalies from becoming a consecutive-run pattern.","command":"python -m pytest -q tests/test_analysis.py::test_isolated_anomalies_do_not_form_second_order_pattern"},{"id":"positional-episodes","description":"Use integer intervals for position-only episode evidence.","command":"python -m pytest -q tests/test_analysis.py::test_positional_episode_uses_integer_interval"}],"manualChecks":["Inspect observations and triggering_samples for examples/spike.json and distinguish episode grouping from pattern findings."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","criteria":[{"criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"covered","evidence":["src/anomalyzer/recipes.py:344","src/anomalyzer/recipes.py:388","tests/test_analysis.py:441"],"notes":"The injected spike retains its observed value, expectation, standardized residual, and point trigger."},{"criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"covered","evidence":["src/anomalyzer/baseline.py:31","src/anomalyzer/recipes.py:174","tests/test_analysis.py:441","tests/test_trend.py:36"],"notes":"Assertions establish repeatability and unchanged earlier evidence under later evaluation changes with fixed seasonal settings. Automatic positional season discovery has the separately documented streaming limitation."},{"criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"covered","evidence":["src/anomalyzer/evidence.py:95","tests/test_analysis.py:274"],"notes":"The test checks one three-sample run, exact contributing positions, direction, and first detection."},{"criterion":"Separated point departures remain separate episodes, and positional episode intervals use integer coordinates.","status":"covered","evidence":["src/anomalyzer/evidence.py:12","src/anomalyzer/evidence.py:23","tests/test_analysis.py:263","tests/test_analysis.py:369"],"notes":"Separated spikes produce two episodes without a consecutive-run pattern; positional evidence produces interval [9, 9]."}],"ignore_findings":[],"done":true,"input_fingerprint":"062833cc1bd19d5734b0f3c63d1e81666ffb568ead3399e359d62db2c0c0677d","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Find spikes and drops"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","label":"Know when the expected bas...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","acceptance":["Repeating seasonal structure with stable calibration can yield supported readiness.","Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","Insufficient calibration support yields not_assessed, including when the run otherwise completes.","Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment."],"guidance":["Readiness does not prove real-world effectiveness and is not an automatic action-policy gate."],"path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/readiness.py","src/anomalyzer/recipes.py","tests/test_readiness.py"],"automatedChecks":[{"id":"supported-reference","description":"Support repeated seasonality and preserve readiness under later changes.","command":"python -m pytest -q tests/test_readiness.py::test_repeating_season_and_trend_support_reference"},{"id":"weak-season","description":"Warn on unstable seasonality even when no findings occur.","command":"python -m pytest -q tests/test_readiness.py::test_unstable_declared_season_warns_without_calling_series_inapplicable tests/test_readiness.py::test_documented_random_example_has_no_findings_but_warns_on_reference"},{"id":"calibration-drift","description":"Warn when residual calibration drifts despite a repeated training season.","command":"python -m pytest -q tests/test_readiness.py::test_calibration_drift_warns_even_when_training_season_repeats"},{"id":"not-assessed","description":"Distinguish uncalibrated history and too few calibration samples.","command":"python -m pytest -q tests/test_readiness.py::test_readiness_is_not_assessed_without_calibrated_evaluation tests/test_readiness.py::test_short_calibration_is_not_assessed_even_when_analysis_completes"}],"manualChecks":["Read readiness reasons alongside findings; a lack of findings alone does not establish a trustworthy reference."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","criteria":[{"criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"covered","evidence":["src/anomalyzer/readiness.py:128","tests/test_readiness.py:8"],"notes":"Assertions establish repeated seasonality, strong calibration improvement, and supported readiness."},{"criterion":"Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","status":"covered","evidence":["src/anomalyzer/readiness.py:113","src/anomalyzer/readiness.py:122","tests/test_readiness.py:28","tests/test_readiness.py:41"],"notes":"Completed analyses retain caution reasons for weak seasonal repetition or calibration-center drift."},{"criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"covered","evidence":["src/anomalyzer/readiness.py:77","tests/test_readiness.py:56","tests/test_readiness.py:63"],"notes":"Both uncalibrated history and a completed run with three calibration samples return not_assessed."},{"criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"covered","evidence":["src/anomalyzer/recipes.py:327","tests/test_readiness.py:8"],"notes":"Readiness reads training/calibration inputs once; tests compare identical segment and aggregate assessments after a later shift."}],"ignore_findings":[],"done":true,"input_fingerprint":"d01367627e058ea39c78ee0e3616061a2e0725d06e8acd0b84b46a9655431e57","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Assess baseline quality"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","label":"Follow evolving growth whi...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","acceptance":["Each rolling expectation uses only observations before the sample being scored.","Persistent growth changes can be followed more closely than by the frozen default recipe.","An isolated extreme value is held out of future model references rather than creating a seasonal echo.","Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected."],"guidance":["A persistent third extreme miss in one direction may be admitted to the rolling reference; this is not a manual regime reset."],"path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"causal-window","description":"Keep a known rolling expectation invariant when future values are appended.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_expectation_is_one_step_and_prefix_invariant"},{"id":"changing-growth","description":"Follow a persistent growth change and report model admissions.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_tracks_a_persistent_growth_change"},{"id":"isolated-extreme","description":"Exclude an isolated extreme from later reference estimates.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_contains_an_isolated_extreme"},{"id":"bounded-settings","description":"Reject conflicting or unbounded adaptive settings.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_settings_reject_conflicting_or_unbounded_trend"}],"manualChecks":["Inspect the adaptive recipe's rolling-window parameters and frozen calibration before interpreting its final expectation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","criteria":[{"criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"covered","evidence":["src/anomalyzer/baseline.py:61","tests/test_analysis.py:48"],"notes":"All rolling-reference positions precede the scored index. The independent expectation remains unchanged after future values are appended."},{"criterion":"Persistent growth changes can be followed more closely than by the frozen default recipe.","status":"covered","evidence":["src/anomalyzer/recipes.py:145","src/anomalyzer/recipes.py:370","tests/test_analysis.py:58"],"notes":"The test requires adaptive held-out MAE below half the frozen recipe's MAE and a close final expectation."},{"criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","tests/test_analysis.py:81"],"notes":"The isolated spike is excluded from model references; its next seasonal expectation closely matches the control. Persistent third misses are admitted as guidance explicitly permits."},{"criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:41","tests/test_analysis.py:104"],"notes":"Both incompatible explicit trend selection and excessive slope lookback raise specific validation errors."}],"ignore_findings":[],"done":true,"input_fingerprint":"984422f1fd3ab3d22cf7b4ba9d0bd06629fc228a8e53d7ad44439cd8d35ec0db","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Follow evolving growth"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","label":"Account for daily and week...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"planned","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","acceptance":["Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","Insufficient support for a requested cycle produces an explicit limitation or declared fallback rather than a silently fitted component.","Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle."],"guidance":["Begin with a bounded daily-and-weekly use case; overlapping cycles share information and cannot be treated as independent evidence.","Proposed automated checks cover combined ordinary cycles, future isolation, unsupported cycle history, and extreme-value influence across cycles.","Manual review should inspect a weekday morning spike and verify the baseline explains ordinary calendar variation without explaining away the spike."],"path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Represent more than one repeating cycle"},"impact":0,"gaps":3,"r":47,"x":170,"y":406},{"id":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","label":"Limit how extreme values d...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","acceptance":["Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","Training and calibration diagnostics identify excluded extreme observations.","An extreme evaluated sample remains evidence while its expected value is used in future references.","Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior."],"guidance":["Protection keeps the observed evidence; it does not delete observations or certify that a departure was an incident."],"path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/robust.py","src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py","tests/test_cli.py"],"automatedChecks":[{"id":"robust-reference","description":"Limit an extreme weekly reference and use earlier references at lag one.","command":"python -m pytest -q tests/test_analysis.py::test_robust_weekly_reference_limits_one_high_week tests/test_analysis.py::test_robust_lag_one_uses_prior_seasonal_values"},{"id":"training-calibration","description":"Record training and calibration exclusions and retain a useful growth fit.","command":"python -m pytest -q tests/test_trend.py::test_robust_training_fit_and_calibration_exclusions"},{"id":"include-choice","description":"Compare robust and include modes through the CLI and show their differing triggers.","command":"python -m pytest -q tests/test_cli.py::test_outlier_handling_cli"}],"manualChecks":["Inspect excluded_from_model, reference_action, and outlier-handling diagnostics alongside the original observed value."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","criteria":[{"criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"covered","evidence":["src/anomalyzer/baseline.py:39","tests/test_analysis.py:25"],"notes":"Assertions check a independently computed weighted reference and invariance when an already-trimmed high value becomes more extreme."},{"criterion":"Training and calibration diagnostics identify excluded extreme observations.","status":"covered","evidence":["src/anomalyzer/trend.py:178","src/anomalyzer/recipes.py:287","tests/test_trend.py:65"],"notes":"Tests identify exact training/calibration exclusions and compare their influence on trend selection and residual scale."},{"criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","src/anomalyzer/recipes.py:392","tests/test_analysis.py:263","tests/test_trend.py:11"],"notes":"The observed departure and trigger survive; only reference_values receives the prior expectation, preventing the next seasonal echo."},{"criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"covered","evidence":["src/anomalyzer/recipes.py:151","src/anomalyzer/recipes.py:475","tests/test_cli.py:95"],"notes":"CLI assertions distinguish robust triggers {65} from include-mode triggers {65, 72}."}],"ignore_findings":[],"done":true,"input_fingerprint":"183b8696fdc9786878b3e54c4a9e5f68180b47e157528bef953274a3487d49a7","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Protect later comparisons"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","label":"Choose the repeating cycle...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","acceptance":["An explicit season length is honored and reported as explicit.","A repeating positional series can infer a seasonal lag and report the selection window.","A positional series without a sufficiently stable candidate falls back to lag one and records why.","The reserved calibration and evaluation tail does not determine positional season selection."],"guidance":["Candidate discovery uses the input prefix before the reserved tail and confirms the candidate in an early window. It is not fully prefix-invariant streaming discovery.","Supply season_length explicitly when stable streaming season selection is required."],"path":".capabilities/analysis/expectations/select-season-length.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/seasonality.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"declared-season","description":"Honor a separate explicit season setting.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"inference-fallback","description":"Infer a repeating lag or record lag-one fallback when no season is supported.","command":"python -m pytest -q tests/test_analysis.py::test_positional_season_inference_and_fallback"},{"id":"reserved-tail","description":"Keep selection unchanged when only the reserved tail changes.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_season_inference_excludes_reserved_tail"},{"id":"cli-inference","description":"Expose the inferred lag and its source for bare JSON values in the CLI.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"}],"manualChecks":["Review selection_window and ranked candidates for a positional series before treating its inferred season as an ongoing monitoring assumption."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","criteria":[{"criterion":"An explicit season length is honored and reported as explicit.","status":"covered","evidence":["src/anomalyzer/recipes.py:55","tests/test_analysis.py:234"],"notes":"Explicit field presence bypasses inference, and the test checks the reported source."},{"criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"covered","evidence":["src/anomalyzer/seasonality.py:96","tests/test_analysis.py:247"],"notes":"The repeating seven-position series selects lag 7 and reports confirmation window [0, 27]."},{"criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"covered","evidence":["src/anomalyzer/seasonality.py:79","src/anomalyzer/seasonality.py:84","tests/test_analysis.py:247"],"notes":"Candidate-strength and early-confirmation failures return lag 1 with a reason; the quadratic example verifies fallback."},{"criterion":"The reserved calibration and evaluation tail does not determine positional season selection.","status":"covered","evidence":["src/anomalyzer/seasonality.py:69","tests/test_capability_behaviors.py:147"],"notes":"Discovery excludes calibration_size plus one reserved sample. Changing that tail preserves the complete inference diagnostics; arbitrary appended-prefix invariance is explicitly outside the guidance."}],"ignore_findings":[],"done":true,"input_fingerprint":"3317b1727dc3cabfcfc9b87808d9249d0d59f111b8fd5f8a1fad653b21343980","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Choose a repeating cycle"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","label":"Interpret rate changes wit...","area":"analysis","scopes":["analysis","analysis/rates"],"scope":"analysis/rates","scopeLabel":"analysis/rates","status":"planned","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","acceptance":["Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","Zero denominators remain unavailable, and low support is explicitly marked; intervals remain within the valid zero-to-one rate domain.","Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations."],"guidance":["Independence and stable exposure are modeling assumptions; clustered events, overdispersion, or changing populations require explicit limitations.","Proposed automated checks cover zero denominators, tiny counts, zero-or-all-success boundaries, and equal rates at different volumes.","Manual review should compare sparse and high-volume conversion changes and confirm the explanation discloses assumptions without inventing incident probability."],"path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Judge rates using their exposure"},"impact":0,"gaps":3,"r":47,"x":194,"y":524},{"id":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","label":"See changes before the day...","area":"analysis","scopes":["analysis","analysis/resolution"],"scope":"analysis/resolution","scopeLabel":"analysis/resolution","status":"implemented","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","acceptance":["Callers can select sum, mean, or last aggregation for completed and partial periods.","Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","An explicitly unfinished source sample is excluded from both analysis views.","A source cadence that does not evenly divide 24 hours is rejected."],"guidance":["Periods are fixed 24-hour windows, not daylight-saving-aware local calendar days.","Native and daily views share source observations and are correlated.","Legacy schema 1.0 requires complete input and does not support resets. Schema 1.1 labels sparse native/completed-day views and preserves reset boundaries; daily sums require declared additive volume and fully observed children."],"path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","dependencies":["data/quality/exclude-incomplete-periods","analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"selected-aggregation","description":"Check sum, mean, and last for completed days and trailing partial aggregates.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_multi_resolution_uses_selected_aggregation"},{"id":"two-views","description":"Return completed days, native evidence, and a linked trailing partial-day marker.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_current_period_and_keeps_two_views"},{"id":"unfinished-source","description":"Exclude a marked incomplete source sample from both views.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days"},{"id":"dividing-cadence","description":"Reject subperiod durations that do not evenly divide a day.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_rejects_non_dividing_frequency"}],"manualChecks":["Inspect the partial aggregate's observed_aggregate and intraday_evidence_refs together with the two method namespaces."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","criteria":[{"criterion":"Callers can select sum, mean, or last aggregation for completed and partial periods.","status":"covered","evidence":["src/anomalyzer/contracts.py:18","src/anomalyzer/recipes.py:539","tests/test_capability_behaviors.py:178"],"notes":"Parameterized assertions check independently known completed and partial values for all three aggregation choices."},{"criterion":"Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","status":"covered","evidence":["src/anomalyzer/recipes.py:617","src/anomalyzer/recipes.py:667","src/anomalyzer/recipes.py:680","tests/test_analysis.py:170"],"notes":"Separate datasets, configurations, and namespaces preserve the two correlated views."},{"criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"covered","evidence":["src/anomalyzer/recipes.py:639","src/anomalyzer/recipes.py:690","tests/test_analysis.py:170"],"notes":"The marker contains finalized progress, observed aggregate, native evidence references, and an explicit exclusion flag."},{"criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"covered","evidence":["src/anomalyzer/validation.py:44","src/anomalyzer/recipes.py:564","tests/test_analysis.py:200"],"notes":"Both child datasets use prepared finalized observations; tests verify counts and retention of the excluded original value."},{"criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"covered","evidence":["src/anomalyzer/recipes.py:572","tests/test_analysis.py:221"],"notes":"A five-hour cadence raises the specific dividing-cadence error."}],"ignore_findings":[],"done":true,"input_fingerprint":"cdcdea1166a4e3d2523b594bec938142135e17a6b4279732e52819a486d1b4f2","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Compare finalized subperiods"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","label":"Compare days and weeks in ...","area":"data","scopes":["data","data/calendar"],"scope":"data/calendar","scopeLabel":"data/calendar","status":"planned","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","acceptance":["Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","Periods report UTC boundaries and local labels; daylight-saving days can contain 23 or 25 hours without being forced into fixed 24-hour windows.","Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation."],"guidance":["Calendar grouping needs a distinct contract from fixed elapsed cadence; it must not silently reinterpret existing frequency or aggregation settings.","Proposed automated checks cover spring transitions, fall transitions, week boundaries, and missing native slots in a calendar period.","Manual review should inspect a daylight-saving week and confirm local labels, observed coverage, and completed-versus-open comparisons are understandable."],"path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Declare local calendar boundaries"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","label":"Analyze many entities with...","area":"data","scopes":["data","data/cohorts"],"scope":"data/cohorts","scopeLabel":"data/cohorts","status":"planned","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","acceptance":["Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","Training, calibration, relationships, and cases do not cross entity boundaries unless a separate explicit aggregate comparison is requested.","The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident."],"guidance":["Batch orchestration should preserve existing per-request contracts rather than silently lifting schema limits; multiple-testing limits stay visible.","Proposed automated checks cover partition isolation, batch-versus-single equivalence, one-entity failure, and shared resource-budget exhaustion.","Manual review should inspect a mixed batch of healthy, changed, and sparse entities and confirm the summary does not equate missing support with health."],"path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Keep each entity's evidence separate"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","label":"Know when missing updates ...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"planned","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","acceptance":["Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","Results distinguish current, late, missing, and not-assessed freshness and expose the last observed update and the deadline used.","A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision."],"guidance":["Freshness depends on declared collection expectations; it does not prove source completeness and does not start a scheduler or send a notification.","Proposed automated checks cover exact deadlines, allowed lateness, absent source updates, and stale-source propagation to relationships.","Manual review should compare a stale feed with a true metric drop and confirm the report distinguishes collection uncertainty from numerical departure."],"path":".capabilities/data/quality/report-source-freshness.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Check whether the sources are current"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","label":"See how late or corrected ...","area":"data","scopes":["data","data/revisions"],"scope":"data/revisions","scopeLabel":"data/revisions","status":"planned","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","acceptance":["A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","Results preserve prior and new evidence versions and identify findings that were added, changed, or withdrawn with traceable source lineage.","Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results."],"guidance":["Schema-1.1 snapshot revision selection and as-of availability are implemented in data/quality/analyze-sparse-observations; durable prior/new evidence diffing and automatic reconciliation in this capability remain planned.","A corrected event-time result does not replace the record of what was known earlier; callers control retention and any notification about a revision.","Proposed automated checks cover late completion, corrected calibration data, downstream relationship changes, and conflicting source versions.","Manual review should inspect a drop withdrawn after ingestion completes and confirm both the original limitation and the revision explanation are retained."],"path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Reconcile revisions without losing history"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","label":"Choose settings using held...","area":"evaluation","scopes":["evaluation","evaluation/comparisons"],"scope":"evaluation/comparisons","scopeLabel":"evaluation/comparisons","status":"planned","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","acceptance":["Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","The output preserves every candidate's settings and evaluation provenance; adopting a configuration requires an explicit caller decision."],"guidance":["Bounded comparison is not unrestricted parameter search; held-out results describe the supplied history and do not guarantee future effectiveness.","Proposed automated checks cover split isolation, comparable replay inputs, failed candidate retention, and unchanged active settings after comparison.","Manual review should compare an earlier but noisier configuration with a quieter one and confirm the tradeoff is visible without an unsupported winner."],"path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Review the tradeoff before adopting settings"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","label":"Measure detection timing a...","area":"evaluation","scopes":["evaluation","evaluation/outcomes"],"scope":"evaluation/outcomes","scopeLabel":"evaluation/outcomes","status":"planned","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","acceptance":["Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","Timing reports distinguish the incident's supplied start from the first available candidate, calibrated detection, and policy-eligible evidence.","Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown."],"guidance":["Detection timing does not measure response effectiveness or establish that an unmatched finding was false; incident labels are caller judgments.","Proposed automated checks cover boundary matching, ambiguous overlaps, incomplete label coverage, and separate candidate versus eligibility timing.","Manual review should inspect a matched and an unmatched episode and confirm the reported timing and accuracy limits agree with the supplied label coverage."],"path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Compare detection with reviewed outcomes"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","label":"Replay history as it would...","area":"evaluation","scopes":["evaluation","evaluation/replay"],"scope":"evaluation/replay","scopeLabel":"evaluation/replay","status":"planned","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","acceptance":["Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","Repeating a replay preserves numerical evidence and first-detection timing apart from generated IDs and runtime; stopped replays retain completed steps."],"guidance":["Event-time replay does not reconstruct arrival-time availability unless the caller supplies versioned arrival history.","Proposed automated checks cover future-tail isolation, short-history cutoffs, deterministic repeated replay, and interrupted replay retention.","Manual review should inspect one change from its first available observation through calibration and confirm the explanation uses only then-known evidence."],"path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Reconstruct the evidence available at the time"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","label":"Update evidence when the n...","area":"execution","scopes":["execution","execution/incremental"],"scope":"execution/incremental","scopeLabel":"execution/incremental","status":"planned","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","acceptance":["Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","Supported explicitly configured recipes produce the same numerical evidence as full batch analysis of the combined history, including cross-chunk patterns.","Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations."],"guidance":["Start with explicit season lengths; input-prefix season discovery is not currently guaranteed to be invariant when the history grows.","Proposed automated checks cover chunk-versus-batch equivalence, cross-chunk CUSUM, idempotent retry, and failed-append state preservation.","Manual review should follow a sustained change across several append batches and confirm detection timing and provisional evidence remain understandable."],"path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Analyze the next finalized samples"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","label":"Keep useful evidence when ...","area":"execution","scopes":["execution","execution/resilience"],"scope":"execution/resilience","scopeLabel":"execution/resilience","status":"implemented","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","acceptance":["A method failure preserves evidence completed before the failure and reports the error.","Runtime budgets are checked cooperatively and produce an explicit stop reason.","CLI byte limits and execution failures produce distinct error exits.","A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial."],"guidance":["Default limits are 10,000 source observations per request, 10 MB CLI input, and 60 seconds cooperative runtime.","Numerical results are deterministic for fixed declared settings; generated run IDs and runtimes vary."],"path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/budget.py","src/anomalyzer/recipes.py","src/anomalyzer/orchestrator.py","src/anomalyzer/io.py","src/anomalyzer/cli.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"method-and-budget","description":"Retain pre-failure evidence and report a runtime-budget stop.","command":"python -m pytest -q tests/test_analysis.py::test_failure_and_budget"},{"id":"cli-limits","description":"Enforce byte limits and distinguish execution failure in CLI output and exit status.","command":"python -m pytest -q tests/test_cli.py::test_size_limit_and_execution_exit"},{"id":"partial-siblings","description":"Preserve healthy analysis when a sibling dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Inspect run status, stop_reason, and each target's status before relying on a result; confirm limits are presented as cooperative rather than hard process timeouts."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","criteria":[{"criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"covered","evidence":["src/anomalyzer/recipes.py:400","src/anomalyzer/recipes.py:448","tests/test_analysis.py:509"],"notes":"The failure regression preserves evidence through index 44 and reports the injected error at index 45."},{"criterion":"Runtime budgets are checked cooperatively and produce an explicit stop reason.","status":"covered","evidence":["src/anomalyzer/budget.py:18","src/anomalyzer/recipes.py:108","src/anomalyzer/recipes.py:508","tests/test_analysis.py:509"],"notes":"Cooperative target/sample checks expose runtime_budget and budget_exceeded; this is not a hard process timeout."},{"criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:429","tests/test_cli.py:108"],"notes":"The full suite executed the byte-limit and execution-exit regression successfully; invalid input exits 2 and execution failure exits 1."},{"criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:125","src/anomalyzer/orchestrator.py:171","src/anomalyzer/orchestrator.py:185","tests/test_capability_behaviors.py:159","tests/test_capability_behaviors.py:30"],"notes":"Healthy dataset evidence survives target-specific failure; inspected incomplete-usage reports partial with no invented healthy relationship."}],"ignore_findings":[],"done":true,"input_fingerprint":"146a9c8396fd7a2d039f75077dc0b82a0778685d20c3782aa284d5f0b6ed94be","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect incomplete execution"},"impact":0,"gaps":0,"r":38,"x":720,"y":642},{"id":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","label":"Resume local analysis afte...","area":"execution","scopes":["execution","execution/state"],"scope":"execution/state","scopeLabel":"execution/state","status":"planned","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","acceptance":["A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","Restoring a supported state and continuing yields the same numerical evidence and detection positions as uninterrupted processing of the same observations.","Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled."],"guidance":["Stored observations and context can contain sensitive data; document exactly what a snapshot contains and avoid automatic persistence.","Proposed automated checks cover restart equivalence, corrupt snapshots, version or settings mismatch, and output overwrite protection.","Manual review should inspect a snapshot and resumed explanation and confirm their provenance and retained-data scope are clear to an application owner."],"path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Preserve state across restarts"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","label":"Understand findings in a r...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","acceptance":["Point-anomaly episode counts and residual-pattern findings are labeled separately.","Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","Readiness status remains visible, including when it was not assessed."],"guidance":["Readable summaries should retain the evidence's claim limits and allow drilldown into JSON."],"path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/cli.py","tests/test_cli.py","examples/agentic/conversion.json","examples/agentic/unit-cost.json"],"automatedChecks":[{"id":"episodes-patterns","description":"Distinguish shift patterns from point episodes in readable output.","command":"python -m pytest -q tests/test_cli.py::test_readable_output_distinguishes_episodes_from_patterns"},{"id":"relationship-case","description":"Explain a conversion case with entity, units, comparison, and a suggested check.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_relationship_request_through_cli"},{"id":"explicit-rule","description":"Describe the unit-cost rule boundary and unassessed readiness.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the conversion and unit-cost example output and confirm the descriptions state the supported numerical comparison in plain language."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","criteria":[{"criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"covered","evidence":["src/anomalyzer/cli.py:347","src/anomalyzer/cli.py:405","tests/test_cli.py:143"],"notes":"Readable tests distinguish zero-episode residual shifts from a spike episode and retain the overlap warning."},{"criterion":"Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","status":"covered","evidence":["src/anomalyzer/cli.py:291","src/anomalyzer/cli.py:308","src/anomalyzer/cli.py:331","tests/test_cli.py:117","examples/agentic/conversion.json"],"notes":"Conversion output identifies site=demo, observed/expected orders_per_visit, the relationship, and a collection-completeness investigation question."},{"criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"covered","evidence":["src/anomalyzer/cli.py:318","tests/test_cli.py:134","examples/agentic/unit-cost.json"],"notes":"Unit-cost output names review-above-13c and its [0, 0.13] boundary as a declared rule crossing."},{"criterion":"Readiness status remains visible, including when it was not assessed.","status":"covered","evidence":["src/anomalyzer/cli.py:352","src/anomalyzer/cli.py:399","tests/test_cli.py:134"],"notes":"Readable output includes not_assessed and its supporting reasons without claiming measured alert accuracy."}],"ignore_findings":[],"done":true,"input_fingerprint":"98108868b000972fd9b3c3f082846fbd3f3b08c399ef7ed424c23245d6935fef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Read the findings"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","label":"Share an investigation tha...","area":"interfaces","scopes":["interfaces","interfaces/export"],"scope":"interfaces/export","scopeLabel":"interfaces/export","status":"planned","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","acceptance":["Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","Source observations and context are included only by explicit selection; omitted data and resulting limits on independent reproduction are disclosed."],"guidance":["Fingerprints identify included provenance; they do not prove source truth. A portable report must make privacy scope and reproduction limits inspectable.","Proposed automated checks cover manifest consistency, partial-run export, omitted-source disclosure, and overwrite protection.","Manual review should open a bundle outside the original session and confirm a reviewer can trace one finding and see exactly what cannot be reproduced."],"path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Preserve a portable evidence record"},"impact":0,"gaps":3,"r":47,"x":927.3333333333334,"y":760},{"id":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","label":"Install a versioned analys...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","acceptance":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"guidance":["Guard owns scheduling, persistence, business severity, incidents and notifications.","A published package does not establish integration into UsageTap Guard."],"path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","interfaces/cli/analyze-files-and-streams","interfaces/mcp/connect-read-only-local-server"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["pyproject.toml","src/anomalyzer/_version.py","scripts/release.py","scripts/publish_release.py","scripts/verify_distribution.py","scripts/verify_lambda.sh","examples/library/guard_smoke.py",".github/workflows/ci.yml",".github/workflows/release.yml","docs/releases/README.md","tests/test_release.py"],"automatedChecks":[{"id":"release-safety","description":"Reject mismatched tags, altered assets and mutable publication.","command":"python -m pytest -q tests/test_release.py"},{"id":"installed-artifacts","description":"Install wheel and sdist outside the checkout and invoke public APIs.","command":"python scripts/verify_distribution.py --python 3.12"},{"id":"release-inventory","description":"Check archive contents, version metadata and SHA-256 inventory.","command":"python scripts/release.py verify"},{"id":"linux-offline-consumer","description":"Bundle and run with networking disabled in Linux Python 3.12.","command":"bash scripts/verify_lambda.sh"}],"manualChecks":["Follow the release guide and confirm Guard retains domain severity and incident ownership; inspect readiness and evidence limits before adopting."],"verificationGaps":[{"code":"declared-gap","message":"interfaces/python/install-versioned-library: First immutable GitHub Release requires owner setup and a reviewed version tag.","capabilityId":"interfaces/python/install-versioned-library","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\python\\install-versioned-library.capability.yaml"}],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for interfaces/python/install-versioned-library.","criteria":[{"criterion":"Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","status":"partial","evidence":["src/anomalyzer/_version.py:1","scripts/release.py:1","scripts/release.py:20","scripts/release.py:39","scripts/release.py:48","scripts/release.py:54","scripts/publish_release.py:19","scripts/publish_release.py:27","scripts/publish_release.py:29","scripts/publish_release.py:39","scripts/verify_lambda.sh:14","scripts/verify_lambda.sh:17","scripts/verify_lambda.sh:19","scripts/verify_lambda.sh:25","scripts/verify_lambda.sh:27",".github/workflows/ci.yml:14",".github/workflows/ci.yml:19",".github/workflows/ci.yml:27",".github/workflows/ci.yml:41",".github/workflows/ci.yml:46",".github/workflows/release.yml:28","docs/releases/README.md:12","docs/releases/README.md:13","docs/releases/README.md:14","docs/releases/README.md:17","docs/releases/README.md:18","tests/test_release.py:60","tests/test_release.py:71","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","status":"partial","evidence":["scripts/verify_distribution.py:1","scripts/verify_distribution.py:42","examples/library/guard_smoke.py:32","examples/library/guard_smoke.py:49","examples/library/guard_smoke.py:65","examples/library/guard_smoke.py:69","examples/library/guard_smoke.py:75",".github/workflows/ci.yml:57","docs/releases/README.md:13","docs/releases/README.md:208","docs/releases/README.md:211","docs/releases/README.md:213"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:34","scripts/release.py:81","scripts/release.py:82","scripts/release.py:85","scripts/release.py:92","scripts/publish_release.py:1","scripts/publish_release.py:25","scripts/publish_release.py:26","scripts/publish_release.py:27","scripts/publish_release.py:28","scripts/verify_distribution.py:41","scripts/verify_lambda.sh:24","examples/library/guard_smoke.py:39",".github/workflows/ci.yml:42",".github/workflows/ci.yml:45",".github/workflows/ci.yml:46",".github/workflows/ci.yml:49",".github/workflows/ci.yml:51",".github/workflows/release.yml:31",".github/workflows/release.yml:33",".github/workflows/release.yml:36",".github/workflows/release.yml:37","docs/releases/README.md:3","docs/releases/README.md:12","docs/releases/README.md:14","docs/releases/README.md:15","docs/releases/README.md:19","tests/test_release.py:1","tests/test_release.py:21","tests/test_release.py:23","tests/test_release.py:52","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:81","scripts/release.py:82","scripts/publish_release.py:40","scripts/verify_distribution.py:41","scripts/verify_lambda.sh:6","scripts/verify_lambda.sh:17","scripts/verify_lambda.sh:19",".github/workflows/ci.yml:30",".github/workflows/ci.yml:31",".github/workflows/ci.yml:51",".github/workflows/ci.yml:52",".github/workflows/ci.yml:90","docs/releases/README.md:12","docs/releases/README.md:15","docs/releases/README.md:58","docs/releases/README.md:59","docs/releases/README.md:75"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","partial: Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","partial: Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","partial: A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","name":"Python integration","label":"Python integration / Embed analysis in my application / Install a released dependency"},"impact":0,"gaps":1,"r":41,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","label":"Ask an agent to explain un...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","acceptance":["The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit."],"guidance":["This contract covers the supplied instructions and demonstrated analysis workflow; it does not guarantee that every agent host follows the instructions.","Synthetic examples demonstrate mechanics, not real-world detection accuracy."],"path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/agents/openai.yaml","src/anomalyzer/cli.py","src/anomalyzer/mcp.py","tests/test_cli.py","tests/test_agentic.py","tests/test_mcp.py"],"automatedChecks":[{"id":"structured-cli","description":"Exercise the documented structured CLI workflow and preserve incomplete-history evidence.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence tests/test_cli.py::test_bare_json_values_cli"},{"id":"declared-relationships","description":"Check explicit ratio semantics and unavailable source evidence.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"mcp-drilldown","description":"Follow compact analysis to a finding and its source evidence.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case"}],"manualChecks":["Follow the conversion example through the CLI and MCP drilldown; compare the Skill instructions with actual evidence and check that its proposed text separates calibrated findings, early candidates, readiness, and caller-owned actions."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","criteria":[{"criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:13",".agents/skills/anomaly-detection/SKILL.md:41","tests/test_cli.py:13"],"notes":"Executable resolution and JSON commands are explicit. Manual conversion and spike CLI runs use the installed numerical engine."},{"criterion":"The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:24",".agents/skills/anomaly-detection/SKILL.md:32","tests/test_cli.py:51","tests/test_agentic.py:211"],"notes":"Instructions preserve scope, units, missingness, declared relationships, and exact alignment; existing runtime checks establish these prerequisites."},{"criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:101",".agents/skills/anomaly-detection/SKILL.md:134",".agents/skills/anomaly-detection/agents/openai.yaml:4"],"notes":"The supplied instructions lead with plain-language changes and then qualify maturity/readiness/overlap. This establishes instructional content, not measured agent compliance."},{"criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:54",".agents/skills/anomaly-detection/SKILL.md:174","src/anomalyzer/mcp.py:169","tests/test_mcp.py:54"],"notes":"The documented three-tool workflow matches executed conversion drilldown, with session-only run IDs and explicit read-only/caller-owned delivery limits."}],"ignore_findings":[],"done":true,"input_fingerprint":"751723214aa340b1227ec347c22f130749fcdbfd49327c21b4c27ef68d7f0341","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Get an evidence-grounded explanation"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","label":"See the evidence in a char...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","acceptance":["The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes."],"guidance":["Native chart rendering and interactive figure features depend on the agent host; the checked-in fallback produces a static SVG for a single dataset.","A chart explains structured analysis; it does not replace the detector result."],"path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/scripts/quick_chart.py",".agents/skills/anomaly-detection/references/visualization.md","tests/test_skill_chart.py"],"automatedChecks":[{"id":"observed-evidence","description":"Render real observed and expected data and only actual point triggers.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_preserves_real_point_evidence"},{"id":"maturity-and-resets","description":"Keep provisional evidence non-triggering and label reviewed regimes.","command":"python -m pytest -q tests/test_skill_chart.py::test_provisional_scores_are_not_charted_as_triggers tests/test_skill_chart.py::test_quick_chart_marks_reviewed_resets"},{"id":"unsupported-scope","description":"Reject a multi-dataset result rather than mixing units.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_rejects_multiple_datasets"},{"id":"output-protection","description":"Require an explicit overwrite for existing charts.","command":"python -m pytest -q tests/test_skill_chart.py::test_chart_cli_requires_explicit_overwrite"}],"manualChecks":["Inspect a spike fallback chart and the detailed figure instructions; verify that axes, initialization context, trigger markers, and interpretation limits are clear, and that the Skill requests a chart only when the user asks."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","criteria":[{"criterion":"The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:189",".agents/skills/anomaly-detection/SKILL.md:226"],"notes":"The Skill explicitly requires a visual request after analysis and separates quick versus detailed rendering instructions."},{"criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:81",".agents/skills/anomaly-detection/scripts/quick_chart.py:59","tests/test_skill_chart.py:28","tests/test_skill_chart.py:43","tests/test_skill_chart.py:54"],"notes":"Real-result tests establish observed/expected lines, trigger-only markers, provisional non-triggering, units, initialization shading, and reset boundaries; the spike chart was inspected visually."},{"criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:85",".agents/skills/anomaly-detection/scripts/quick_chart.py:251","tests/test_skill_chart.py:63","tests/test_skill_chart.py:68"],"notes":"Multiple resolved datasets raise a specific error. The CLI test preserves an existing chart and replaces it only with explicit overwrite."},{"criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/references/visualization.md:31",".agents/skills/anomaly-detection/references/visualization.md:37",".agents/skills/anomaly-detection/references/visualization.md:45",".agents/skills/anomaly-detection/references/visualization.md:86"],"notes":"Detailed instructions require index alignment, maturity distinctions, first detection timing, visible unavailable status, and qualified causal claims; native interactive rendering remains host-dependent."}],"ignore_findings":[],"done":true,"input_fingerprint":"2780b6c3e363f89a30a0709772c0a46a9a96f1567fabb27e63934a3ca80dc982","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Inspect a requested chart"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","label":"Find which segments contri...","area":"investigation","scopes":["investigation","investigation/cohorts"],"scope":"investigation/cohorts","scopeLabel":"investigation/cohorts","status":"planned","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","acceptance":["Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","Segment ranking retains absolute signed contributions and baseline support; an arithmetic contributor is not automatically labeled anomalous or causal."],"guidance":["Ratios and overlapping customer groups are not additive totals; separate segment expectations must not be assumed to sum to a fitted total expectation.","Proposed automated checks cover exact reconciliation, offsetting segment changes, missing segments, and zero total departure.","Manual review should inspect a regional decline with an offsetting increase and confirm the explanation preserves signed contributions and evidence limits."],"path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Locate contributors to a total change"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","label":"Keep reviewed outcomes bes...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","acceptance":["A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","Review export distinguishes analyst judgment from detector output; annotation alone does not change calibration, eligibility, or regime boundaries."],"guidance":["Caller-supplied reviewer identity is attribution, not built-in authentication; storage, access control, response execution, and incident ownership stay with callers.","Proposed automated checks cover evidence-version linking, unresolved reviews, revision retention, and unchanged numerical evidence after annotation.","Manual review should inspect a disputed case with two review revisions and confirm judgments, uncertainties, and original findings remain distinguishable."],"path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Record the reviewed outcome"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","label":"Find earlier reviewed case...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","acceptance":["Search is bounded to caller-selected records and explicit compatibility filters; it does not scan unrelated local files or require a model API.","Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match."],"guidance":["Similar historical evidence does not establish the same cause or authorize repeating a past response; caller-selected records may be incomplete or biased.","Proposed automated checks cover compatibility filtering, bounded corpus access, missing comparison context, and no-match behavior.","Manual review should compare superficially similar cases with different outcomes and confirm precedent is presented as investigation context."],"path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Inspect comparable reviewed history"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","label":"Change action eligibility ...","area":"investigation","scopes":["investigation","investigation/policy"],"scope":"investigation/policy","scopeLabel":"investigation/policy","status":"implemented","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","acceptance":["Default policy leaves action and notification eligibility disabled.","Explicit-rule and calibrated-departure eligibility are controlled independently.","Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged."],"guidance":["Eligibility does not send notifications, execute actions, check readiness, or confer authority to act.","Monitoring schedules, delivery, reviewed incident state, and deduplication belong to the caller."],"path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","dependencies":["investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/agent.py","src/anomalyzer/cases.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"independent-notifications","description":"Replay notification and action choices without changing numerical evidence or the original result.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_separates_action_from_notification"},{"id":"calibrated-choice","description":"Enable calibrated departures only with the corresponding policy choice.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_enables_calibrated_departures_only_when_requested"},{"id":"nontriggering-candidates","description":"Keep descriptive candidates ineligible even under a permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"}],"manualChecks":["Inspect the unit-cost example under default and allow_explicit_rules policies and compare eligibility with the unchanged numerical violation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","criteria":[{"criterion":"Default policy leaves action and notification eligibility disabled.","status":"covered","evidence":["src/anomalyzer/contracts.py:329","tests/test_capability_behaviors.py:74"],"notes":"Default flags disable both eligibility types, including after replay resets policy."},{"criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"covered","evidence":["src/anomalyzer/agent.py:47","tests/test_capability_behaviors.py:123"],"notes":"Explicit-only permission does not enable a calibrated departure; the matching calibrated choice does."},{"criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"covered","evidence":["src/anomalyzer/agent.py:53","tests/test_capability_behaviors.py:101","tests/test_capability_behaviors.py:134"],"notes":"Notification-only replay enables a met rule while keeping action disabled; permissive settings still cannot enable candidates."},{"criterion":"Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged.","status":"covered","evidence":["src/anomalyzer/agent.py:42","src/anomalyzer/cases.py:57","src/anomalyzer/cases.py:105","tests/test_capability_behaviors.py:101","tests/test_agentic.py:49"],"notes":"Assertions compare unchanged original serialization, methods, evidence references, stable IDs, and eligibility-dependent revision changes."}],"ignore_findings":[],"done":true,"input_fingerprint":"ee67bcc1ec1380b8d84a0556d838d648ae6cef3f4f9b9e34e1b71e90eb00aeae","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Review action eligibility"},"impact":0,"gaps":0,"r":38,"x":1086.6666666666667,"y":878},{"id":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","label":"Review a possible new base...","area":"investigation","scopes":["investigation","investigation/regimes"],"scope":"investigation/regimes","scopeLabel":"investigation/regimes","status":"planned","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","acceptance":["A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","Producing a proposal leaves the active regime and original findings unchanged.","A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect."],"guidance":["A proposed boundary is not a confirmed cause or a retrospectively available detection; adaptive reference updates remain a separate behavior.","Proposed automated checks cover unchanged active results, short new regimes, rejected proposals, and accepted explicit-reset equivalence.","Manual review should inspect a persistent outage-like drop and a known launch and confirm neither becomes ordinary behavior merely because it persists."],"path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Review a proposed regime transition"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","label":"Distinguish changing perfo...","area":"relationships","scopes":["relationships","relationships/cohorts"],"scope":"relationships/cohorts","scopeLabel":"relationships/cohorts","status":"planned","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","acceptance":["Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","The report preserves aggregate and cohort rates, population weights, and the declared decomposition formula and reference-period convention.","Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes."],"guidance":["Decomposition is descriptive accounting; it does not estimate an intervention effect or establish that the comparison populations are causally equivalent.","Proposed automated checks cover stable rates with changing mix, stable mix with changing rates, offsetting cohort rates, and new or empty cohorts.","Manual review should inspect a Simpson's-paradox example and confirm the aggregate movement and cohort evidence can both be understood correctly."],"path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Separate performance changes from population mix"},"impact":0,"gaps":3,"r":47,"x":1270,"y":878},{"id":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","label":"Check whether a metric fol...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","acceptance":["A supplied lag and coefficient use the predictor at the declared earlier sample position.","Derived values retain response and predictor source indexes and source cutoffs in lineage.","When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","Samples lacking the earlier predictor or awaiting coefficient training remain unavailable."],"guidance":["Lag is measured in samples and is never searched automatically.","Exactly one of coefficient or training_size is required."],"path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","dependencies":["relationships/quality/require-compatible-sources"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_examples.py","examples/agentic/lagged-response.json"],"automatedChecks":[{"id":"supplied-response","description":"Use the declared lag and coefficient with correct predictor indexes.","command":"python -m pytest -q tests/test_agentic.py::test_lagged_response_uses_supplied_coefficient_and_prior_sample"},{"id":"frozen-coefficient","description":"Keep fitted coefficient and earlier lineage invariant when future samples arrive.","command":"python -m pytest -q tests/test_agentic.py::test_estimated_lag_coefficient_is_frozen_and_prefix_invariant"},{"id":"delayed-example","description":"Run the checked-in lagged-response demonstration against expected results.","command":"python -m pytest -q tests/test_examples.py::test_agentic_demo_checks_all_cases"},{"id":"pending-training","description":"Retain unavailable lineage until the full declared training prefix arrives, then freeze its coefficient.","command":"python -m pytest -q tests/test_agentic.py::test_lag_fit_waits_for_complete_training_prefix"}],"manualChecks":["Inspect the lagged-response example's leading unavailable values, coefficient parameters, and first usable derived sample."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","criteria":[{"criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"covered","evidence":["src/anomalyzer/relationships.py:115","tests/test_agentic.py:254"],"notes":"Supplied lag/coefficient uses the declared earlier predictor position and preserves the expected response residual."},{"criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"covered","evidence":["src/anomalyzer/relationships.py:152","tests/test_agentic.py:254","examples/agentic/lagged-response.json"],"notes":"Available and pending lineage retains existing source indexes and timestamp cutoffs; missing predictors do not become zero."},{"criterion":"When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","status":"covered","evidence":["src/anomalyzer/relationships.py:47","tests/test_agentic.py:294","tests/test_agentic.py:330"],"notes":"The coefficient is fitted once after the complete declared prefix; the independent 20/7 result and prefix-invariance assertions pass."},{"criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"covered","evidence":["src/anomalyzer/relationships.py:122","tests/test_agentic.py:330"],"notes":"Short prefixes retain null derived values and explicit unavailable reasons, while an unidentifiable completed fit has a visible error."}],"ignore_findings":[],"done":true,"input_fingerprint":"fc4516ef6ab903dfa4918268d97049aff04fbf9556620ccb8259ff830ba85785","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Check delayed responses"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878},{"id":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","label":"Inspect related metrics be...","area":"relationships","scopes":["relationships","relationships/resolution"],"scope":"relationships/resolution","scopeLabel":"relationships/resolution","status":"implemented","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","acceptance":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"guidance":["Daily ratios and differences are supported. Daily lagged, normalized-residual and joint-condition definitions remain explicitly inapplicable; their native evidence remains eligible. Local-calendar aggregation is separate work.","Initial support should use explicit relationships and compatible boundaries; cross-cadence joining or automatic resampling needs a separate contract.","Proposed automated checks cover ratio-of-totals arithmetic, missing sources, mismatched period boundaries, and open-period evidence separation.","Manual review should inspect a conversion decline within an unfinished day and verify the explanation separates native evidence from partial totals."],"path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","dependencies":["data/quality/analyze-sparse-observations","relationships/comparisons/compare-declared-metrics"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","src/anomalyzer/relationships.py","src/anomalyzer/agent.py","tests/test_sparse.py","examples/sparse/unfinished_daily.json","docs/sparse-data.md"],"automatedChecks":[{"id":"ratio-of-sums","description":"Compute daily ratios from compatible sums and preserve missing-child coverage.","command":"python -m pytest -q tests/test_sparse.py::test_complete_daily_ratio_divides_sums_and_missing_child_invalidates_only_its_view"},{"id":"view-policies","description":"Keep native/daily rules separate and preserve case identity on replay.","command":"python -m pytest -q tests/test_sparse.py::test_daily_rules_are_not_silently_reused_from_hourly_rules tests/test_sparse.py::test_multi_resolution_mcp_drilldown_selects_correct_view_and_replays_case_ids"},{"id":"view-limits","description":"Report unsupported daily lag semantics without discarding native evidence.","command":"python -m pytest -q tests/test_sparse.py::test_unsupported_daily_relationship_is_explicit_and_does_not_reinterpret_lag"}],"manualChecks":["Inspect progression chapter 22 and confirm incomplete progress is not presented as a completed total or independent corroboration."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for relationships/resolution/compare-finalized-subperiods.","criteria":[{"criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:22","src/anomalyzer/sparse_views.py:139","src/anomalyzer/sparse_views.py:143","src/anomalyzer/sparse_views.py:164","src/anomalyzer/agent.py:14","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:23","src/anomalyzer/agent.py:29","tests/test_sparse.py:18","tests/test_sparse.py:59","tests/test_sparse.py:87","tests/test_sparse.py:99","tests/test_sparse.py:117","docs/sparse-data.md:31","docs/sparse-data.md:127","docs/sparse-data.md:159","docs/sparse-data.md:199","docs/sparse-data.md:203"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:122","src/anomalyzer/sparse_views.py:123","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:36","src/anomalyzer/agent.py:34","tests/test_sparse.py:10","tests/test_sparse.py:40","tests/test_sparse.py:50","tests/test_sparse.py:60","tests/test_sparse.py:62","docs/sparse-data.md:17","docs/sparse-data.md:41","docs/sparse-data.md:68","docs/sparse-data.md:97","docs/sparse-data.md:99"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:70","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:170","src/anomalyzer/relationships.py:90","src/anomalyzer/relationships.py:173","tests/test_sparse.py:230","tests/test_sparse.py:259","tests/test_sparse.py:313","docs/sparse-data.md:41","docs/sparse-data.md:111","docs/sparse-data.md:194","docs/sparse-data.md:197","docs/sparse-data.md:198"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:130","src/anomalyzer/sparse_views.py:159","src/anomalyzer/relationships.py:157","src/anomalyzer/relationships.py:192","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:47","tests/test_sparse.py:217","tests/test_sparse.py:247","tests/test_sparse.py:251","tests/test_sparse.py:264","tests/test_sparse.py:386","examples/sparse/unfinished_daily.json:208","docs/sparse-data.md:8","docs/sparse-data.md:11","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:45"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","partial: Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","partial: Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","partial: Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","name":"Related-metric analysis","label":"Related-metric analysis / Monitor the periods my business uses / Compare related metrics while the period is open"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878}],"links":[{"source":"data/inputs/accept-analysis-requests","target":"analysis/evidence/calibrate-departures","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/cli/analyze-files-and-streams","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"analysis/expectations/model-seasonal-level-and-growth","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"interfaces/python/analyze-through-public-api","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/identify-residual-patterns","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"investigation/assessments/describe-early-changes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"relationships/quality/require-compatible-sources","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/expectations/reset-declared-regimes","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"data/quality/exclude-incomplete-periods","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/mcp/connect-read-only-local-server","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/rules/check-declared-numerical-rules","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/report-point-anomaly-episodes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/evidence/assess-reference-readiness","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/adapt-to-changing-growth","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/protect-reference-from-extremes","kind":"committed"},{"source":"data/quality/exclude-incomplete-periods","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"execution/resilience/retain-evidence-under-failure-and-limits","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/mcp/connect-read-only-local-server","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/visualize-analysis-on-request","kind":"committed"},{"source":"investigation/rules/check-declared-numerical-rules","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/evaluate-delayed-response","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"},{"source":"relationships/comparisons/compare-declared-metrics","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"}],"scopes":["analysis","analysis/detection","analysis/evidence","analysis/expectations","analysis/rates","analysis/resolution","data","data/calendar","data/cohorts","data/inputs","data/quality","data/revisions","evaluation","evaluation/comparisons","evaluation/outcomes","evaluation/replay","execution","execution/incremental","execution/resilience","execution/state","interfaces","interfaces/cli","interfaces/export","interfaces/mcp","interfaces/python","interfaces/skill","investigation","investigation/assessments","investigation/cases","investigation/cohorts","investigation/history","investigation/policy","investigation/regimes","investigation/rules","relationships","relationships/cohorts","relationships/comparisons","relationships/quality","relationships/resolution"],"storyMaps":["Anomaly detection Skill","Basic CLI anomaly detection","Evidence and review policy","Future - Business-aware expectations","Future - Calendar and data readiness","Future - Cohort investigation","Future - Continuous analysis","Future - Historical evaluation","Future - Investigation memory","Incomplete-period monitoring","Python integration","Read-only MCP server","Related-metric analysis","Trend and seasonal analysis"],"hasUnassignedStoryMap":false};
+const graph = {"nodes":[{"id":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","label":"Bring my metrics into an a...","area":"data","scopes":["data","data/inputs"],"scope":"data/inputs","scopeLabel":"data/inputs","status":"implemented","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","acceptance":["Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","Published schema 1.0 and 1.1 request and result schemas match the public Python contracts."],"guidance":["Null observations remain unknown unless a schema-1.1 additive-volume policy explicitly authorizes confirmed empty collection as zero. Limits apply to the request as a whole.","Describe current supported behavior; historical plans do not add accepted inputs."],"path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["analysis/evidence/calibrate-departures","analysis/expectations/model-seasonal-level-and-growth","data/quality/analyze-sparse-observations","data/quality/identify-distinct-events","execution/resilience/retain-evidence-under-failure-and-limits","interfaces/python/analyze-through-public-api","investigation/assessments/describe-early-changes"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_agentic.py","schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json"],"automatedChecks":[{"id":"input-forms","description":"Analyze a bare list and validate a canonical request and result.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings tests/test_analysis.py::test_constant_and_schema"},{"id":"explicit-scope","description":"Reject schema 1.1 datasets without entity scope.","command":"python -m pytest -q tests/test_agentic.py::test_schema_11_requires_explicit_entity_scope"},{"id":"invalid-inputs","description":"Reject unsupported settings and nonnumeric or nonfinite observations.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_config tests/test_analysis.py::test_invalid_values"},{"id":"published-schemas","description":"Compare all four checked-in schemas with their public contracts.","command":"python -m pytest -q tests/test_analysis.py::test_checked_in_schemas tests/test_agentic.py::test_v11_checked_in_schemas_exist"}],"manualChecks":["Review the input examples in README.md and confirm they describe the two supported schema versions and the four-dataset/four-relationship limit."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","criteria":[{"criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"covered","evidence":["src/anomalyzer/recipes.py:21","src/anomalyzer/contracts.py:96","tests/test_analysis.py:234"],"notes":"Bare-list settings are validated; schema 1.0 limits datasets to exactly one."},{"criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"covered","evidence":["src/anomalyzer/contracts.py:240","src/anomalyzer/contracts.py:335","tests/test_agentic.py:192"],"notes":"Entity scope is required, relationship kinds are discriminated, and references must name distinct known datasets."},{"criterion":"Unknown settings, unsupported recipes, nonfinite values, booleans, and numeric strings are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:7","src/anomalyzer/contracts.py:71","tests/test_analysis.py:526","tests/test_analysis.py:531"],"notes":"Assertions reject each listed category; missing null values remain a separate accepted representation."},{"criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"covered","evidence":["schemas/request.schema.json","schemas/result.schema.json","schemas/request-1.1.schema.json","schemas/result-1.1.schema.json","tests/test_analysis.py:547","tests/test_agentic.py:317"],"notes":"Tests and independent inspection compare each complete published schema with its contract-generated schema. Python cross-field validators supplement JSON Schema constraints."}],"ignore_findings":[],"done":true,"input_fingerprint":"3546cd8f436812ff11c9352a855926677b55cc6287e40f6a5041bd613173e5b4","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Supply a numeric series"},"impact":7,"gaps":0,"r":73,"x":353.33333333333337,"y":170},{"id":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","label":"Understand how much histor...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","acceptance":["Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","Provisional residual scores can be reported but do not trigger point-anomaly episodes.","Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger."],"guidance":["Evidence strength is not a probability of an incident.","Calibration and expectations are separate concerns; each declared regime calibrates again."],"path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/detection/identify-residual-patterns","analysis/detection/report-point-anomaly-episodes","analysis/evidence/assess-reference-readiness","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","relationships/comparisons/compare-declared-metrics"],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"short-history","description":"Keep reference-only evidence for a short series.","command":"python -m pytest -q tests/test_analysis.py::test_short_series_returns_reference_evidence"},{"id":"provisional-evidence","description":"Report provisional scores without point triggers.","command":"python -m pytest -q tests/test_analysis.py::test_early_evidence_maturity_and_no_provisional_trigger"},{"id":"independent-calibration","description":"Check a known residual center, sample scale, and standardized score.","command":"python -m pytest -q tests/test_analysis.py::test_calibration_reference"},{"id":"constant-reference","description":"Handle zero residual variance without spurious episodes.","command":"python -m pytest -q tests/test_analysis.py::test_constant_and_schema"}],"manualChecks":["Inspect a sample's signal_maturity, calibration_samples, standardized_residual, and triggers together before describing a statistical departure."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","criteria":[{"criterion":"Short series return reference evidence and an insufficient_history status instead of inventing calibrated findings.","status":"covered","evidence":["src/anomalyzer/recipes.py:205","src/anomalyzer/recipes.py:445","tests/test_analysis.py:141"],"notes":"The short-series test checks three reference-only samples, absent standardized scores, and no triggers."},{"criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"covered","evidence":["src/anomalyzer/recipes.py:210","tests/test_analysis.py:123"],"notes":"A provisional score of 2 exceeds the configured point threshold of 1 but produces no trigger or episode."},{"criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"covered","evidence":["src/anomalyzer/recipes.py:231","src/anomalyzer/recipes.py:307","tests/test_analysis.py:111"],"notes":"Calibration uses retained residual mean and sample standard deviation once per regime; the independent numerical example produces the expected standardized score."},{"criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"covered","evidence":["src/anomalyzer/contracts.py:104","src/anomalyzer/recipes.py:350","src/anomalyzer/recipes.py:388","tests/test_analysis.py:111"],"notes":"Sample fields retain the comparison and support; the point threshold is inspectable in the enclosing method parameters."}],"ignore_findings":[],"done":true,"input_fingerprint":"ccdc1b2e92f3e647d5a9e6127cc3c4b10bf0da8f1f78af3efb805daa9b1e4305","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Establish evidence support"},"impact":6,"gaps":0,"r":68,"x":170,"y":170},{"id":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","label":"Analyze local files from t...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","acceptance":["The methods command lists supported analysis recipes and residual detectors.","Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status."],"guidance":["CSV callers must select one entity and unit before analysis.","Exit 0 includes completed analysis with anomalies, insufficient history, or inapplicability; exit 2 means invalid input, and exit 1 means execution failure or partial execution."],"path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/python/install-versioned-library","interfaces/skill/guide-anomaly-investigation","interfaces/skill/visualize-analysis-on-request"],"implementationReferences":["src/anomalyzer/cli.py","src/anomalyzer/io.py","src/anomalyzer/__main__.py","tests/test_cli.py"],"automatedChecks":[{"id":"file-stdin","description":"Analyze equivalent file and stdin requests.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence"},{"id":"csv-precedence","description":"Validate selected CSV fields and apply flag-over-file-over-embedded precedence.","command":"python -m pytest -q tests/test_cli.py::test_csv_and_precedence"},{"id":"value-list","description":"Analyze a bare JSON array and expose inferred settings.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"},{"id":"output-and-exits","description":"Protect existing output and distinguish invalid options from failed execution.","command":"python -m pytest -q tests/test_cli.py::test_output_protection tests/test_cli.py::test_methods_and_invalid_options tests/test_cli.py::test_size_limit_and_execution_exit"}],"manualChecks":["Run an example in readable and JSON modes and confirm stdout contains results while diagnostics go to stderr."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","criteria":[{"criterion":"The methods command lists supported analysis recipes and residual detectors.","status":"covered","evidence":["src/anomalyzer/cli.py:426","src/anomalyzer/registry.py:15","tests/test_cli.py:27"],"notes":"Discovery tests check all three recipes and ten residual detectors, with structured stdout and no successful-run diagnostics on stderr."},{"criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:239","tests/test_cli.py:13","tests/test_mcp.py:163"],"notes":"The passing file/stdin regression compares request fingerprints and short-history semantics; cross-interface tests compare complete stable numerical output."},{"criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"covered","evidence":["src/anomalyzer/io.py:33","tests/test_cli.py:51"],"notes":"The passing CSV regression covers timestamped missing cells, positional data, and varying unmapped entity rejection."},{"criterion":"Settings resolve in embedded-request, settings-file, then CLI-flag precedence; bare JSON values can use CLI configuration.","status":"covered","evidence":["src/anomalyzer/cli.py:253","tests/test_cli.py:51","tests/test_cli.py:71"],"notes":"The passing file/flag/embedded precedence regression and bare-values regression establish the merge order and validated configuration."},{"criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"covered","evidence":["src/anomalyzer/io.py:69","src/anomalyzer/cli.py:424","tests/test_cli.py:41","tests/test_cli.py:108"],"notes":"Output protection and exit regressions passed, including explicit overwrite; successful manual examples have empty stderr and incomplete-usage exits 1."}],"ignore_findings":[],"done":true,"input_fingerprint":"86e701ccc0a1aefd407f11b4547784fef497fa1b49ddf6bb69db0d08f757dfdc","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Run from files or stdin"},"impact":4,"gaps":0,"r":58,"x":903.3333333333334,"y":170},{"id":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","label":"Spot sustained shifts and ...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","acceptance":["Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","Trend, oscillation, and mixture findings retain their specific pattern meanings.","Moving-range findings identify variation changes separately from location changes."],"guidance":["Pattern findings may overlap and are not independent confirmations.","A residual location shift does not establish a change in the full distribution."],"path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations","data/quality/identify-distinct-events","investigation/cases/group-related-evidence"],"implementationReferences":["src/anomalyzer/evidence.py","src/anomalyzer/recipes.py","tests/test_analysis.py"],"automatedChecks":[{"id":"location-shift","description":"Find moderate increases and decreases without point anomalies.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_patterns_describe_location_shift_without_point_anomaly"},{"id":"accumulating-change","description":"Detect a known CUSUM accumulation at the expected position.","command":"python -m pytest -q tests/test_analysis.py::test_cusum_detects_accumulated_moderate_location_shift"},{"id":"diagnostic-patterns","description":"Distinguish residual trend, oscillation, and mixture rules.","command":"python -m pytest -q tests/test_analysis.py::test_nelson_diagnostics_have_specific_semantics"},{"id":"variation","description":"Describe a moving-range increase as variation rather than location.","command":"python -m pytest -q tests/test_analysis.py::test_moving_range_detects_short_term_variation_increase"}],"manualChecks":["Compare examples/location_shift_up.json with its point episodes and pattern findings, allowing for overlapping residual rules."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","criteria":[{"criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"covered","evidence":["src/anomalyzer/evidence.py:118","src/anomalyzer/evidence.py:189","tests/test_analysis.py:293"],"notes":"Tests assert no point triggers or episodes and exact upward/downward detection positions for Nelson rules and CUSUM."},{"criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"covered","evidence":["src/anomalyzer/evidence.py:211","src/anomalyzer/evidence.py:230","tests/test_analysis.py:350"],"notes":"The known accumulation detects at index 16 and reports peak CUSUM approximately 6, threshold 5, and increase direction."},{"criterion":"Trend, oscillation, and mixture findings retain their specific pattern meanings.","status":"covered","evidence":["src/anomalyzer/evidence.py:70","src/anomalyzer/evidence.py:122","tests/test_analysis.py:340"],"notes":"Separate predicates and descriptions preserve the three meanings and explicitly limit location/distribution claims."},{"criterion":"Moving-range findings identify variation changes separately from location changes.","status":"covered","evidence":["src/anomalyzer/evidence.py:234","tests/test_analysis.py:360"],"notes":"Adjacent standardized-residual differences produce variation_shift findings; the test checks magnitude, detection index, and interpretation."}],"ignore_findings":[],"done":true,"input_fingerprint":"63716eb67db6b73964b79897d8c81f2891df2586ff213c95ea81c5699b8abbea","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Find sustained residual changes"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","label":"Account for ordinary seaso...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","acceptance":["A supplied season length selects earlier matching seasonal observations for the reference.","Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected."],"guidance":["Expectations support scoring; they are not a standalone forecasting API."],"path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["analysis/expectations/protect-reference-from-extremes","analysis/expectations/reset-declared-regimes","analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py"],"automatedChecks":[{"id":"seasonal-reference","description":"Check independent lagged-reference values.","command":"python -m pytest -q tests/test_analysis.py::test_forecast_independent_reference"},{"id":"growth-and-departures","description":"Keep clean linear and compound growth ordinary while identifying injected departures.","command":"python -m pytest -q tests/test_trend.py::test_trend_and_anomalies"},{"id":"frozen-fit","description":"Keep the training fit stable when later observations change.","command":"python -m pytest -q tests/test_trend.py::test_frozen_fit_and_legacy_mode"},{"id":"explicit-trends","description":"Accept supported trend choices and handle short, constant, and signed series.","command":"python -m pytest -q tests/test_trend.py::test_explicit_modes_and_validation tests/test_trend.py::test_short_history_constant_and_signed_linear"}],"manualChecks":["Compare examples/growth_up.json and examples/compound.json with their selected trend and evidence expectations."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","criteria":[{"criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"covered","evidence":["src/anomalyzer/baseline.py:8","src/anomalyzer/baseline.py:31","tests/test_analysis.py:21"],"notes":"Independent lag values and robust matching-phase assertions establish the seasonal reference calculation."},{"criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"covered","evidence":["src/anomalyzer/trend.py:120","src/anomalyzer/trend.py:144","tests/test_trend.py:11"],"notes":"Positive and negative linear/exponential cases select the expected model, match post-training expectations, and produce no ordinary-growth episodes."},{"criterion":"The fitted trend remains frozen after training within a regime; later changed values do not rewrite that fit.","status":"covered","evidence":["src/anomalyzer/recipes.py:162","tests/test_trend.py:36"],"notes":"Trend fitting occurs at the training boundary; the test asserts identical trend diagnostics and earlier evidence after later observations change."},{"criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:20","tests/test_trend.py:49","tests/test_trend.py:57"],"notes":"Tests exercise disabled, linear, exponential, and invalid quadratic modes, including disclosed short-history fallback."}],"ignore_findings":[],"done":true,"input_fingerprint":"8e9a0361b02b7d3f651366e5fa0675cccb3949e44de448d2ae0d43e6710bd408","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Account for growth and seasonality"},"impact":3,"gaps":0,"r":53,"x":170,"y":170},{"id":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","label":"Embed local analysis in my...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","acceptance":["analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","analyze_series preserves supplied settings and event-time context for a single dataset.","analyze_relationships preserves the declared multi-dataset relationship semantics.","Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime."],"guidance":["All interfaces use local analysis; there is no hosted service or model API requirement."],"path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence","interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/__init__.py","src/anomalyzer/agent.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"request-parity","description":"Match canonical request results across Python, CLI, and full-result MCP compatibility calls.","command":"python -m pytest -q tests/test_mcp.py::test_canonical_request_matches_python_cli_and_mcp tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"series-context","description":"Preserve a single-series event-time cutoff and context across interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp tests/test_mcp.py::test_series_wrapper_preserves_omitted_and_explicit_season_settings"},{"id":"relationship-parity","description":"Match declared relationship evidence across the public interfaces.","command":"python -m pytest -q tests/test_mcp.py::test_relationship_semantics_match_python_cli_and_mcp"}],"manualChecks":["Review the Python API examples in README.md and confirm callers can inspect the returned validated models and JSON."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","criteria":[{"criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"covered","evidence":["src/anomalyzer/__init__.py:3","src/anomalyzer/recipes.py:21","tests/test_analysis.py:234"],"notes":"Bare values and separate settings validate before dispatch; canonical schema 1.0 and 1.1 requests use their respective contracts."},{"criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"covered","evidence":["src/anomalyzer/agent.py:11","tests/test_mcp.py:191","tests/test_mcp.py:291"],"notes":"Field-set handling preserves omitted versus explicit settings and context in the single-series wrapper, verified across dictionary and Settings inputs."},{"criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"covered","evidence":["src/anomalyzer/agent.py:27","tests/test_mcp.py:217"],"notes":"Declared relationships validate and use the shared engine without changing units, lineage, or primitive semantics."},{"criterion":"Equivalent requests retain numerical results across Python, CLI, and full-result MCP compatibility calls apart from run IDs and runtime.","status":"covered","evidence":["tests/test_mcp.py:163","tests/test_mcp.py:191","tests/test_mcp.py:217","src/anomalyzer/__main__.py:1"],"notes":"Passing real CLI subprocess, Python, and full-result MCP parity tests compare stable numerical results apart from identity and runtime."}],"ignore_findings":[],"done":true,"input_fingerprint":"214bd10bb3aa5fad0f74938fe2ef09853a593cdb5b93cba49ec851a628d8ab65","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","name":"Python integration","label":"Python integration / Embed analysis in my application / Call the public API"},"impact":3,"gaps":0,"r":53,"x":903.3333333333334,"y":170},{"id":"investigation/cases/group-related-evidence","title":"Investigate related findings together","label":"Investigate related findin...","area":"investigation","scopes":["investigation","investigation/cases"],"scope":"investigation/cases","scopeLabel":"investigation/cases","status":"implemented","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","acceptance":["Evidence for different declared entities remains in separate cases even at the same event time.","Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","A derived relationship adds context without counting as an additional independent source for severity.","Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection."],"guidance":["Warning and critical are deterministic numerical grouping rules, not validated incident impact.","Diagnostic trend, oscillation, mixture, and variation findings remain distinct from adaptive location-shift cases."],"path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","dependencies":["investigation/assessments/describe-early-changes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["interfaces/cli/explain-findings-in-readable-output","interfaces/mcp/investigate-findings-and-evidence","investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/cases.py","src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py"],"automatedChecks":[{"id":"entity-scope","description":"Separate cases for different entities at the same event time.","command":"python -m pytest -q tests/test_agentic.py::test_cases_are_scoped_by_entity_at_the_same_event_time"},{"id":"source-severity","description":"Escalate numerical severity when two source datasets meet criteria.","command":"python -m pytest -q tests/test_agentic.py::test_two_source_datasets_escalate_case_severity"},{"id":"correlated-context","description":"Keep relationship evidence from double-counting source severity.","command":"python -m pytest -q tests/test_agentic.py::test_relationship_evidence_explains_but_does_not_double_count_severity"},{"id":"first-detection","description":"Group an adaptive location shift before point triggers without using a later pattern peak.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_location_shift_creates_one_early_case_without_point_trigger tests/test_agentic.py::test_location_case_severity_does_not_use_future_pattern_peak"}],"manualChecks":["Inspect a case's contributing targets, explanation, evidence references, and investigation questions; distinguish it from a point-anomaly episode."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","criteria":[{"criterion":"Evidence for different declared entities remains in separate cases even at the same event time.","status":"covered","evidence":["src/anomalyzer/cases.py:42","tests/test_agentic.py:69"],"notes":"Entity participates in grouping and stable identity; the test requires separate cases and distinct IDs at the same time."},{"criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"covered","evidence":["src/anomalyzer/cases.py:72","tests/test_agentic.py:85"],"notes":"Two qualifying source datasets produce critical numerical severity and named source explanations."},{"criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"covered","evidence":["src/anomalyzer/cases.py:72","src/anomalyzer/cases.py:116","tests/test_agentic.py:100"],"notes":"The independent-source count includes dataset criteria only. Derived evidence remains explanatory and can still meet the separately declared magnitude rule."},{"criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"covered","evidence":["src/anomalyzer/early.py:193","src/anomalyzer/early.py:237","src/anomalyzer/orchestrator.py:115","tests/test_agentic.py:122","tests/test_agentic.py:146"],"notes":"Tests require one moderate-shift case without point triggers and restrict assessment references/strength to samples at or before first detection."}],"ignore_findings":[],"done":true,"input_fingerprint":"6830af8a2f064123c12adb062f4793e446b97b3b805fd73a934ea84cd0d5d9cf","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Investigate related evidence"},"impact":3,"gaps":0,"r":53,"x":1086.6666666666667,"y":170},{"id":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","label":"Assess sparse observations...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","acceptance":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"guidance":["Default missingness is unknown. A modeled expectation is never a measurement.","Caller business severity, incident state and notifications remain outside the engine.","Optional event-identity-v1 IDs retain simultaneous distinct calls; only a separately asserted trusted sequence permits ordering their measurements."],"path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":["data/quality/identify-distinct-events","relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","src/anomalyzer/orchestrator.py","src/anomalyzer/relationships.py","src/anomalyzer/early.py","tests/test_sparse.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"source-semantics","description":"Preserve unknown periods and require explicit empty-volume authorization.","command":"python -m pytest -q tests/test_sparse.py -k \"zero or omitted or grid or duplicate or expansion\""},{"id":"causal-boundaries","description":"Retain real phase, event times, resets and as-of revision eligibility.","command":"python -m pytest -q tests/test_sparse.py -k \"phase or gap or reset or event or revision or arrival\""},{"id":"public-interfaces","description":"Preserve useful rules, relationships, views, cases and API parity.","command":"python -m pytest -q tests/test_sparse.py -k \"rule or relationship or daily or parity\""},{"id":"progressive-examples","description":"Reproduce the full numbered sequence and sparse worked examples.","command":"python examples/run_demo.py"}],"manualChecks":["Inspect progression figures 14–22 for true gaps, distinguished policy zeros, original event time, unavailable checks and caller-owned business severity."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/analyze-sparse-observations.","criteria":[{"criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","status":"partial","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:71","src/anomalyzer/contracts.py:85","src/anomalyzer/contracts.py:86","src/anomalyzer/sparse.py:3","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:42","src/anomalyzer/sparse.py:48","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse_views.py:74","src/anomalyzer/sparse_views.py:81","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:97","src/anomalyzer/sparse_views.py:119","src/anomalyzer/orchestrator.py:63","src/anomalyzer/orchestrator.py:127","src/anomalyzer/orchestrator.py:139","src/anomalyzer/orchestrator.py:191","src/anomalyzer/relationships.py:36","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:98","src/anomalyzer/relationships.py:131","src/anomalyzer/relationships.py:141","src/anomalyzer/early.py:58","src/anomalyzer/early.py:137","tests/test_sparse.py:44","tests/test_sparse.py:45","tests/test_sparse.py:50","tests/test_sparse.py:51","tests/test_sparse.py:57","docs/sparse-data.md:4","docs/sparse-data.md:29","docs/sparse-data.md:37","docs/sparse-data.md:39","docs/sparse-data.md:40","examples/sparse_examples.py:22","examples/sparse_examples.py:23","examples/sparse_examples.py:25","examples/sparse_examples.py:26","examples/sparse_examples.py:28","examples/render_readme_progression.py:135","examples/render_readme_progression.py:139","examples/render_readme_progression.py:140","examples/render_readme_progression.py:166","examples/render_readme_progression.py:206"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","status":"partial","evidence":["src/anomalyzer/contracts.py:27","src/anomalyzer/contracts.py:49","src/anomalyzer/contracts.py:51","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:56","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:181","src/anomalyzer/sparse.py:187","src/anomalyzer/sparse.py:189","src/anomalyzer/sparse_views.py:15","src/anomalyzer/sparse_views.py:98","src/anomalyzer/sparse_views.py:107","src/anomalyzer/sparse_views.py:108","src/anomalyzer/sparse_views.py:112","src/anomalyzer/orchestrator.py:22","src/anomalyzer/orchestrator.py:23","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:129","src/anomalyzer/orchestrator.py:141","src/anomalyzer/relationships.py:185","src/anomalyzer/relationships.py:189","src/anomalyzer/relationships.py:191","src/anomalyzer/early.py:46","src/anomalyzer/early.py:48","src/anomalyzer/early.py:51","src/anomalyzer/early.py:59","src/anomalyzer/early.py:61","tests/test_sparse.py:20","tests/test_sparse.py:108","tests/test_sparse.py:111","tests/test_sparse.py:113","tests/test_sparse.py:119","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:42","docs/sparse-data.md:106","docs/sparse-data.md:108","examples/sparse_examples.py:16","examples/sparse_examples.py:33","examples/sparse_examples.py:50","examples/sparse_examples.py:51","examples/sparse_examples.py:91","examples/render_readme_progression.py:241","examples/render_readme_progression.py:312","examples/render_readme_progression.py:317","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:242","src/anomalyzer/contracts.py:245","src/anomalyzer/contracts.py:258","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:57","src/anomalyzer/sparse.py:63","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:127","src/anomalyzer/sparse_views.py:155","src/anomalyzer/sparse_views.py:157","src/anomalyzer/orchestrator.py:37","src/anomalyzer/orchestrator.py:38","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:178","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:40","src/anomalyzer/relationships.py:57","src/anomalyzer/relationships.py:58","src/anomalyzer/relationships.py:96","src/anomalyzer/relationships.py:97","src/anomalyzer/early.py:140","src/anomalyzer/early.py:147","tests/test_sparse.py:59","tests/test_sparse.py:67","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:128","docs/sparse-data.md:26","docs/sparse-data.md:96","docs/sparse-data.md:101","docs/sparse-data.md:132","docs/sparse-data.md:142","examples/sparse_examples.py:26","examples/sparse_examples.py:36","examples/sparse_examples.py:38","examples/sparse_examples.py:57","examples/sparse_examples.py:81","examples/render_readme_progression.py:654","examples/render_readme_progression.py:697","examples/render_readme_progression.py:884","examples/render_readme_progression.py:896","examples/render_readme_progression.py:917"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","status":"partial","evidence":["src/anomalyzer/contracts.py:13","src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:297","src/anomalyzer/contracts.py:298","src/anomalyzer/contracts.py:437","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:34","src/anomalyzer/sparse.py:64","src/anomalyzer/sparse.py:66","src/anomalyzer/sparse.py:83","src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:28","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:94","src/anomalyzer/orchestrator.py:106","src/anomalyzer/orchestrator.py:107","src/anomalyzer/orchestrator.py:133","src/anomalyzer/orchestrator.py:166","src/anomalyzer/orchestrator.py:180","src/anomalyzer/relationships.py:160","src/anomalyzer/relationships.py:162","src/anomalyzer/relationships.py:166","tests/test_sparse.py:59","tests/test_sparse.py:75","tests/test_sparse.py:99","tests/test_sparse.py:123","tests/test_sparse.py:180","docs/sparse-data.md:26","docs/sparse-data.md:41","docs/sparse-data.md:69","docs/sparse-data.md:78","docs/sparse-data.md:96","examples/sparse_examples.py:59","examples/sparse_examples.py:95","examples/render_readme_progression.py:686","examples/render_readme_progression.py:699","examples/render_readme_progression.py:710","examples/render_readme_progression.py:767","examples/render_readme_progression.py:884"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","partial: Frozen eligible training and calibration support point references across gaps, while pattern and change comparisons respect gaps and explicit resets.","partial: Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","partial: As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Assess sparse source data"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","label":"Analyze observations in th...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","acceptance":["Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","Calendar duplicate timestamps require an aggregation policy; identified event-index calls can share original timestamps while duplicate logical IDs are rejected and arbitrary tie order supplies no sequence evidence.","An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration."],"guidance":["No interpolation or implicit filling, resampling, or unit conversion occurs. Schema-1.1 source policies may explicitly authorize confirmed empty additive volume as zero.","Duplicate null values propagate through aggregation.","Stable event identity and trusted sequence are separate caller contracts; legacy event inputs without IDs still require unique timestamps."],"path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":["data/quality/exclude-incomplete-periods","relationships/quality/require-compatible-sources"],"implementationReferences":["src/anomalyzer/events.py","tests/test_events.py","docs/sparse-data.md","src/anomalyzer/sparse.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/contracts.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"coordinate-contract","description":"Reject partial timestamp metadata and positional as_of cutoffs.","command":"python -m pytest -q tests/test_analysis.py::test_timestamp_metadata_is_all_or_nothing"},{"id":"local-time","description":"Normalize valid local times and reject ambiguous or nonexistent times.","command":"python -m pytest -q tests/test_analysis.py::test_local_time"},{"id":"duplicates-cutoff","description":"Require an explicit duplicate policy and retain counts after the cutoff.","command":"python -m pytest -q tests/test_analysis.py::test_duplicate_sort_and_cutoff tests/test_events.py::test_distinct_same_time_calls_keep_values_references_and_cases"},{"id":"cadence-context","description":"Check fixed weekly cadence and preserve cutoff semantics across interfaces.","command":"python -m pytest -q tests/test_analysis.py::test_weekly_cadence tests/test_mcp.py::test_single_series_context_matches_python_cli_and_mcp"}],"manualChecks":["Inspect a timestamped result's data_quality transformations, UTC window, and resolved context to confirm preparation is explainable."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","criteria":[{"criterion":"Timestamped datasets provide both timestamps and frequency; positional datasets provide neither.","status":"covered","evidence":["src/anomalyzer/contracts.py:77","tests/test_analysis.py:418"],"notes":"Partial coordinate metadata and positional as_of are rejected."},{"criterion":"Local timestamps require a timezone, and ambiguous or nonexistent daylight-saving times require an explicit offset.","status":"covered","evidence":["src/anomalyzer/validation.py:8","tests/test_analysis.py:500"],"notes":"UTC round trips detect ambiguous/nonexistent local times; valid local time normalizes to the expected UTC hour."},{"criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"covered","evidence":["src/anomalyzer/validation.py:87","src/anomalyzer/validation.py:103","tests/test_analysis.py:478"],"notes":"Explicit mean/sum handling retains duplicate counts and transformation descriptions. Nulls propagate rather than becoming zero."},{"criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"covered","evidence":["src/anomalyzer/validation.py:105","src/anomalyzer/validation.py:111","tests/test_analysis.py:491","tests/test_mcp.py:191"],"notes":"Filtering is inclusive through as_of; cadence compares UTC elapsed timedeltas, including fixed seven-day weeks."}],"ignore_findings":[],"done":true,"input_fingerprint":"791748764cdc0dfe81d987a908179930d858670606e51b64407172295776b261","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Prepare the comparison"},"impact":2,"gaps":0,"r":48,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","label":"Let an agent inspect the e...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","acceptance":["The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","get_evidence exposes the selected finding's numerical support and relationship source lineage.","Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated."],"guidance":["Run results are retained only in the bounded server-session cache; missing or evicted run IDs require reanalysis.","Cases, episodes, and residual patterns may overlap."],"path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":["interfaces/mcp/connect-read-only-local-server","interfaces/skill/guide-anomaly-investigation"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py"],"automatedChecks":[{"id":"case-drilldown","description":"Analyze, page findings, and inspect exact conversion case values and lineage.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case tests/test_mcp.py::test_schema_11_episode_is_paged_and_has_relationship_evidence"},{"id":"unavailable-finding","description":"Expose a quality finding when latest relationship evidence is unavailable.","command":"python -m pytest -q tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"},{"id":"early-candidate","description":"Inspect descriptive candidate evidence without inventing a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case tests/test_mcp.py::test_sparse_insufficient_history_keeps_descriptive_candidate_visible"},{"id":"sample-pagination","description":"Page supporting samples for a residual pattern.","command":"python -m pytest -q tests/test_mcp.py::test_get_evidence_pages_pattern_samples"}],"manualChecks":["Follow analyze, list_findings, and get_evidence for the conversion example and check that the headline is supported by the retrieved values."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","criteria":[{"criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"covered","evidence":["src/anomalyzer/mcp.py:170","src/anomalyzer/mcp_views.py:247","tests/test_mcp.py:54"],"notes":"The conversion manual call returns a relationship headline, preview, counts, readiness, and session-scoped run ID."},{"criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:132","src/anomalyzer/mcp_views.py:260","tests/test_mcp.py:313","tests/test_mcp.py:108"],"notes":"Cases, early candidates, episodes, patterns, and quality findings feed a bounded page; relationship episode regression verifies schema 1.1 pagination."},{"criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:273","src/anomalyzer/mcp_views.py:294","tests/test_mcp.py:54","tests/test_mcp.py:313"],"notes":"Manual drilldown returns observed 8/130 versus expected 0.1 and source indexes orders=14 and qualified_visits=14."},{"criterion":"Unavailable latest evidence and non-triggering early candidates remain visible, while large pattern sample sets are paginated.","status":"covered","evidence":["src/anomalyzer/mcp_views.py:118","src/anomalyzer/mcp_views.py:169","src/anomalyzer/mcp_views.py:337","tests/test_mcp.py:87","tests/test_mcp.py:137","tests/test_mcp.py:341"],"notes":"Passing tests retain short-history and inapplicable-method candidates plus unavailable latest evidence, while pattern samples are paginated."}],"ignore_findings":[],"done":true,"input_fingerprint":"a09b0003780bcd2f9ccf8e3d8ff58b3147c118760b984757d7027b5810fd9bef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Drill into findings and evidence"},"impact":2,"gaps":0,"r":48,"x":927.3333333333334,"y":288},{"id":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","label":"Spot early changes with a ...","area":"investigation","scopes":["investigation","investigation/assessments"],"scope":"investigation/assessments","scopeLabel":"investigation/assessments","status":"implemented","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","acceptance":["The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","Missing values break the comparison history, and zero previous values do not create relative-change evidence.","Early candidates remain non-triggering and ineligible for action even under a permissive policy."],"guidance":["An early candidate does not establish an anomaly, stable baseline, cause, or incident."],"path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":["investigation/cases/group-related-evidence","investigation/rules/check-declared-numerical-rules"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_mcp.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"prefix-invariant-facts","description":"Describe short history causally without creating cases.","command":"python -m pytest -q tests/test_agentic.py::test_short_history_assessments_are_useful_and_prefix_invariant"},{"id":"comparison-boundaries","description":"Reset comparisons at missing observations and avoid dividing by zero.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"candidate-eligibility","description":"Keep candidates ineligible for action and notifications under permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"},{"id":"agent-visibility","description":"Expose a short-history candidate and its claim limits without a case.","command":"python -m pytest -q tests/test_mcp.py::test_short_history_candidate_is_visible_without_a_case"}],"manualChecks":["Inspect a candidate's comparison population, maturity, established claim, and not_established text before describing it."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","criteria":[{"criterion":"The first observation records a value, the second records a change, and later departures from the prior range can be candidates.","status":"covered","evidence":["src/anomalyzer/early.py:73","src/anomalyzer/early.py:96","tests/test_agentic.py:35"],"notes":"Assertions establish observation/change/change/departure_candidate for [10,12,11,20]."},{"criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"covered","evidence":["src/anomalyzer/early.py:50","src/anomalyzer/early.py:141","tests/test_agentic.py:35"],"notes":"Each assessment precedes the prior-history update; tests compare complete serialized assessments for every prefix."},{"criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"covered","evidence":["src/anomalyzer/early.py:53","src/anomalyzer/early.py:58","tests/test_capability_behaviors.py:87"],"notes":"The zero comparison has no relative change; after a missing value, comparison samples reset to zero."},{"criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"covered","evidence":["src/anomalyzer/early.py:105","src/anomalyzer/cases.py:35","tests/test_capability_behaviors.py:134","tests/test_mcp.py:341"],"notes":"Candidates have criterion_met=false and remain action/notification-ineligible after policy replay, without creating cases."}],"ignore_findings":[],"done":true,"input_fingerprint":"debdeeeaea559950f53461a4298cb9c2e3a1681f83a43e8a981ed9b51968a4d0","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect early evidence"},"impact":2,"gaps":0,"r":48,"x":1110.6666666666667,"y":288},{"id":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","label":"Compare the same entity an...","area":"relationships","scopes":["relationships","relationships/quality"],"scope":"relationships/quality","scopeLabel":"relationships/quality","status":"implemented","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","acceptance":["Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","Difference relationships require identical declared source units and matching output units.","An invalid relationship is reported as failed while valid sibling dataset results remain available."],"guidance":["A failed or unavailable relationship does not establish that its metric is normal.","No implicit timestamp matching, interpolation, unit conversion, or zero-filling occurs."],"path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"implementationReferences":["src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"exact-alignment","description":"Reject mismatched timestamps, cadence, entity, duplicates, and positional relationship sources.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relationship_requires_exact_source_alignment"},{"id":"compatible-units","description":"Report a difference unit mismatch without suppressing sibling results.","command":"python -m pytest -q tests/test_agentic.py::test_alignment_unit_and_partial_failure_are_specific"},{"id":"unambiguous-declaration","description":"Reject fields that do not belong to the selected relationship kind.","command":"python -m pytest -q tests/test_agentic.py::test_contract_rejects_ambiguous_relationship_fields"},{"id":"healthy-siblings","description":"Retain a healthy dataset's evidence when another dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Read both target-level statuses and errors before interpreting a partial run; check source dataset errors when a relationship source failed."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","criteria":[{"criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"covered","evidence":["src/anomalyzer/alignment.py:42","tests/test_capability_behaviors.py:30"],"notes":"Parameterized assertions require specific errors for timestamp, cadence, and entity mismatches."},{"criterion":"Relationship sources are timestamped and have unique timestamps; implicit resampling and duplicate aggregation are not accepted for lineage.","status":"covered","evidence":["src/anomalyzer/alignment.py:23","src/anomalyzer/alignment.py:44","tests/test_capability_behaviors.py:30"],"notes":"Positional sources fail exact alignment; even an explicit duplicate-mean setting is rejected for relationship source lineage."},{"criterion":"Difference relationships require identical declared source units and matching output units.","status":"covered","evidence":["src/anomalyzer/relationships.py:41","tests/test_agentic.py:242"],"notes":"Source and output unit checks are explicit. Tests cover mismatched source units and successful matching units."},{"criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:171","tests/test_capability_behaviors.py:30","tests/test_capability_behaviors.py:159"],"notes":"Relationship errors remain target-specific, healthy dataset results survive, and the overall mixed run is partial."}],"ignore_findings":[],"done":true,"input_fingerprint":"06321ab5806a8276cc3954df5d6f4dd7f059f67600038854569787ff51b420d6","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Align comparable sources"},"impact":2,"gaps":0,"r":48,"x":1294,"y":288},{"id":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","label":"Compare against a new base...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","acceptance":["Integer positions and exact observed dates or timestamps can identify reset boundaries.","Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","Repeated, out-of-range, or unmatched reset points are rejected.","CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries."],"guidance":["Reset points are explicit caller decisions. Known-event notes and review-state fields do not rebuild regimes.","Multi-resolution analysis does not support reset_points."],"path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["data/quality/analyze-sparse-observations"],"implementationReferences":["src/anomalyzer/validation.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py"],"automatedChecks":[{"id":"independent-regimes","description":"Retrain separate regimes without cross-boundary evidence or patterns.","command":"python -m pytest -q tests/test_analysis.py::test_manual_resets_retrain_without_crossing_boundaries"},{"id":"invalid-boundaries","description":"Reject invalid or duplicate resolved reset positions.","command":"python -m pytest -q tests/test_analysis.py::test_invalid_manual_resets"},{"id":"cli-boundaries","description":"Resolve position and date reset flags through the CLI.","command":"python -m pytest -q tests/test_cli.py::test_manual_reset_cli"}],"manualChecks":["Inspect the resolved reset positions and segment diagnostics for examples/steps_reset.json."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","criteria":[{"criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"covered","evidence":["src/anomalyzer/validation.py:129","tests/test_analysis.py:381"],"notes":"Positions and observed dates are exercised end to end; timestamp normalization and exact matching are implemented directly."},{"criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"covered","evidence":["src/anomalyzer/recipes.py:119","src/anomalyzer/recipes.py:143","tests/test_analysis.py:381"],"notes":"Each segment initializes separate state and passes its own start to forecasts. Tests verify three calibrated regimes and no cross-boundary evidence or findings."},{"criterion":"Repeated, out-of-range, or unmatched reset points are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:51","src/anomalyzer/validation.py:149","tests/test_analysis.py:413"],"notes":"Validation rejects literal duplicates and different declarations resolving to the same position, invalid positions, and unmatched dates/timestamps."},{"criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"covered","evidence":["tests/test_cli.py:82","src/anomalyzer/recipes.py:53"],"notes":"CLI assertions establish positions [50, 100] and removal of the unmarked-regime episodes."}],"ignore_findings":[],"done":true,"input_fingerprint":"067063f972c5b70d990d6cb0125892731a97a7f482698d7b331a1a30229ac223","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Restart after a reviewed change"},"impact":1,"gaps":0,"r":43,"x":194,"y":288},{"id":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","label":"Avoid mistaking unfinished...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","acceptance":["Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","The observed values and timestamps of unfinished periods remain available in data quality output.","Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","A relationship with an unavailable latest source reports that unavailability rather than normal behavior."],"guidance":["Completeness is declared by the caller; the engine does not infer collection completeness."],"path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","dependencies":["data/quality/prepare-timestamped-series"],"suggestedDependencies":[],"dependents":["analysis/resolution/monitor-within-unfinished-days"],"implementationReferences":["src/anomalyzer/sparse.py","src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/validation.py","src/anomalyzer/alignment.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_mcp.py"],"automatedChecks":[{"id":"unfinished-values","description":"Retain an unfinished value without scoring it.","command":"python -m pytest -q tests/test_analysis.py::test_explicit_incomplete_period_is_reported_and_not_scored"},{"id":"trailing-only","description":"Reject incomplete observations followed by completed observations.","command":"python -m pytest -q tests/test_analysis.py::test_incomplete_periods_must_be_trailing"},{"id":"source-and-agent-views","description":"Exclude a partial source sample from both analysis views and expose unavailable relationship evidence.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days tests/test_mcp.py::test_agent_sees_unavailable_latest_evidence"}],"manualChecks":["Inspect examples/agentic/incomplete-usage.json and confirm the latest headline explains missing completeness and does not imply normality."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicitly incomplete trailing observations are reported as data quality evidence and excluded from completed-series analysis.","criteria":[{"criterion":"Incomplete observations form a trailing suffix after chronological preparation.","status":"covered","evidence":["src/anomalyzer/validation.py:44","tests/test_analysis.py:162"],"notes":"Suffix validation runs after timestamp preparation and rejects a completed observation following an incomplete one."},{"criterion":"The observed values and timestamps of unfinished periods remain available in data quality output.","status":"covered","evidence":["src/anomalyzer/validation.py:50","src/anomalyzer/validation.py:114","tests/test_analysis.py:150"],"notes":"Quality records retain original source index, normalized timestamp, observed value, and incomplete status."},{"criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"covered","evidence":["src/anomalyzer/validation.py:53","src/anomalyzer/recipes.py:564","tests/test_analysis.py:150","tests/test_analysis.py:200"],"notes":"Only finalized prepared values enter either scoring path; assertions verify the last scored sample and both multi-resolution counts."},{"criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"covered","evidence":["src/anomalyzer/alignment.py:35","src/anomalyzer/orchestrator.py:150","tests/test_mcp.py:87"],"notes":"Incomplete aligned sources become unavailable values, and latest relationship state/status plus MCP headline explain the lack of evidence."}],"ignore_findings":[],"done":true,"input_fingerprint":"d82ee816fe89a25232e2589587d31bcadfa81be4188a0e1531f4fbfdf0ebcb5e","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Keep unfinished data visible"},"impact":1,"gaps":0,"r":43,"x":377.33333333333337,"y":288},{"id":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","label":"Give an agent read-only ac...","area":"interfaces","scopes":["interfaces","interfaces/mcp"],"scope":"interfaces/mcp","scopeLabel":"interfaces/mcp","status":"implemented","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","acceptance":["Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","Analysis tool calls return structured content together with compact text.","Invalid tool input and unknown session run IDs return structured errors rather than fabricated results."],"guidance":["Compatibility tool names remain callable but are not advertised.","Legacy clients receive isError tool results; modern clients receive JSON-RPC invalid-input errors.","The server provides analysis and evidence inspection; the caller owns monitoring and operational workflows."],"path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","dependencies":["interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":["interfaces/python/install-versioned-library"],"implementationReferences":["src/anomalyzer/mcp.py","src/anomalyzer/mcp_views.py","tests/test_mcp.py","pyproject.toml"],"automatedChecks":[{"id":"tool-discovery","description":"Advertise the three read-only tools and their schema contracts.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_lists_read_only_tools"},{"id":"connection-protocols","description":"Support legacy stdio initialization and modern stateless discovery.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_stdio_legacy_handshake_and_tool_listing tests/test_mcp.py::test_mcp_modern_discovery_and_stateless_tool_listing"},{"id":"structured-content","description":"Return a compact analysis response with structured content and run ID.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_tool_call_returns_structured_content"},{"id":"invalid-calls","description":"Reject malformed input and unknown session run IDs in both protocols.","command":"python -m pytest -q tests/test_mcp.py::test_mcp_invalid_tool_input_is_a_tool_error tests/test_mcp.py::test_paged_tool_inputs_enforce_schema_in_both_protocols"}],"manualChecks":["Review discovery's interpretation guidance and confirm the client can start anomalyzer-mcp in the installed Python environment."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","criteria":[{"criterion":"Tool discovery advertises analyze, list_findings, and get_evidence as read-only with input and output schemas.","status":"covered","evidence":["src/anomalyzer/mcp.py:63","tests/test_mcp.py:27"],"notes":"Discovery exposes three read-only tools with the same strict argument contracts used at dispatch and advertised output schemas."},{"criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"covered","evidence":["src/anomalyzer/mcp.py:228","tests/test_mcp.py:248","tests/test_mcp.py:268"],"notes":"Legacy initialization and modern stateless discovery tests establish protocol metadata and evidence interpretation guidance."},{"criterion":"Analysis tool calls return structured content together with compact text.","status":"covered","evidence":["src/anomalyzer/mcp.py:162","tests/test_mcp.py:39"],"notes":"The manual conversion call returns a compact headline and structured run ID, counts, readiness, and preview."},{"criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"covered","evidence":["src/anomalyzer/mcp.py:156","src/anomalyzer/mcp.py:257","tests/test_mcp.py:235","tests/test_mcp.py:367"],"notes":"Tests reject invalid cursor/limit types, unknown fields, numeric run IDs, and missing session runs in both protocols without fabricated results."}],"ignore_findings":[],"done":true,"input_fingerprint":"7bd92d238f4f9545372c6c35e78aefea60c585bea0a4a8bb99096ab6cef27f96","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","name":"Read-only MCP server","label":"Read-only MCP server / Give an agent access to local evidence / Connect a local server"},"impact":1,"gaps":0,"r":43,"x":903.3333333333334,"y":406},{"id":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","label":"Find values that break my ...","area":"investigation","scopes":["investigation","investigation/rules"],"scope":"investigation/rules","scopeLabel":"investigation/rules","status":"implemented","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","acceptance":["Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","Absolute and relative changes must exceed their declared maximum to violate a rule.","Relative-change checks require a nonzero previous value and do not bridge missing observations.","Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them."],"guidance":["An exact rule crossing does not establish cause, an incident, or business impact."],"path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","dependencies":["investigation/assessments/describe-early-changes"],"suggestedDependencies":[],"dependents":["investigation/policy/replay-action-eligibility"],"implementationReferences":["src/anomalyzer/early.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"rule-boundaries","description":"Check inclusive range boundaries and strict absolute/relative change thresholds.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_explicit_rules_respect_boundaries"},{"id":"relative-availability","description":"Avoid relative violations across zero or missing values.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_relative_rule_does_not_compare_across_zero_or_missing_values"},{"id":"explicit-not-actionable","description":"Create evidence and a case for an exact rule crossing without enabling action by default.","command":"python -m pytest -q tests/test_agentic.py::test_explicit_rule_is_exact_but_not_actionable_without_policy"},{"id":"readable-rule","description":"Identify the exact unit-cost range rule in CLI output.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the unit-cost example's crossed rule and confirm its text distinguishes the declared bound from statistical detection."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","criteria":[{"criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"covered","evidence":["src/anomalyzer/early.py:22","src/anomalyzer/contracts.py:209","tests/test_capability_behaviors.py:74"],"notes":"Boundary values 10 and 20 are accepted; 9 and 21 violate the declared range."},{"criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"covered","evidence":["src/anomalyzer/early.py:25","tests/test_capability_behaviors.py:74"],"notes":"Parameterized tests distinguish equality to the threshold from strict exceedance for both change rules."},{"criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"covered","evidence":["src/anomalyzer/early.py:29","src/anomalyzer/early.py:53","tests/test_capability_behaviors.py:87"],"notes":"Zero/missing predecessors suppress relative comparisons, and missing values reset comparison history."},{"criterion":"Violations identify the exact declared rule and remain action-ineligible unless policy explicitly permits them.","status":"covered","evidence":["src/anomalyzer/early.py:34","src/anomalyzer/early.py:121","tests/test_agentic.py:49","tests/test_cli.py:134"],"notes":"Violation records retain rule ID/kind/bounds; default eligibility is false and replay requires explicit permission."}],"ignore_findings":[],"done":true,"input_fingerprint":"60d24e84faee9c1e2ae8f9337d17334b37c89305093404da2e84c51526111d0b","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Apply my stated limits"},"impact":1,"gaps":0,"r":43,"x":1086.6666666666667,"y":406},{"id":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","label":"Find changes in conversion...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","acceptance":["Ratios retain the computed value, formula, and original source indexes for each sample.","Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","Derived series can be analyzed with the frozen or adaptive single-resolution recipe."],"guidance":["Relationships are caller-supplied declarations, not discovered or causal relationships.","A difference requires identical declared units; a normalized residual requires a positive supplied scale.","A joint condition is a declared 0/1 comparison, not statistical proof."],"path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","dependencies":["relationships/quality/require-compatible-sources","analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":["relationships/resolution/compare-finalized-subperiods"],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_examples.py"],"automatedChecks":[{"id":"ratio-lineage","description":"Explain a changing rate with the exact source indexes and formula.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"other-comparisons","description":"Check declared differences, normalized residuals, and combined conditions.","command":"python -m pytest -q tests/test_agentic.py::test_remaining_relationship_primitives"},{"id":"unavailable-values","description":"Keep zero-denominator and missing-source observations unavailable.","command":"python -m pytest -q tests/test_agentic.py::test_missing_and_zero_denominator_are_unavailable_not_zero"},{"id":"adaptive-relationship","description":"Use the adaptive recipe for a declared relationship.","command":"python -m pytest -q tests/test_agentic.py::test_adaptive_recipe_runs_for_declared_relationships"}],"manualChecks":["Inspect examples/agentic/conversion.json and distinguish its relationship departure from the individual source metrics."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","criteria":[{"criterion":"Ratios retain the computed value, formula, and original source indexes for each sample.","status":"covered","evidence":["src/anomalyzer/relationships.py:84","src/anomalyzer/relationships.py:152","tests/test_agentic.py:211"],"notes":"Assertions check exact ratio values, formula, and both original source indexes."},{"criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"covered","evidence":["src/anomalyzer/relationships.py:97","src/anomalyzer/relationships.py:134","tests/test_agentic.py:285"],"notes":"Parameterized assertions independently specify difference, normalized residual, all-condition, and any-condition outputs."},{"criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"covered","evidence":["src/anomalyzer/relationships.py:89","src/anomalyzer/relationships.py:142","tests/test_agentic.py:226"],"notes":"Lineage retains null values with missing_source_value/zero_denominator reasons. Explicit zero_denominator=error intentionally fails instead, as documented."},{"criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:87","src/anomalyzer/orchestrator.py:135","tests/test_agentic.py:171","tests/test_examples.py:17"],"notes":"Both supported recipes dispatch derived datasets through the shared single-resolution engine."}],"ignore_findings":[],"done":true,"input_fingerprint":"7fa8f7ffdc0d852ad961b3e1753761d1bd3b00b87b8c001a957ec19e26d068b9","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Compare ratios and metric gaps"},"impact":1,"gaps":0,"r":43,"x":1270,"y":406},{"id":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","label":"Find unusual spikes and dr...","area":"analysis","scopes":["analysis","analysis/detection"],"scope":"analysis/detection","scopeLabel":"analysis/detection","status":"implemented","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","acceptance":["A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","Separated point departures remain separate episodes, and positional episode intervals use integer coordinates."],"guidance":["Detection is a numerical departure from the selected expectation, not a claim about cause or business impact."],"path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/recipes.py","src/anomalyzer/evidence.py","tests/test_analysis.py"],"automatedChecks":[{"id":"point-evidence","description":"Detect an injected spike and retain deterministic earlier numerical evidence.","command":"python -m pytest -q tests/test_analysis.py::test_spike_and_no_lookahead"},{"id":"adjacent-run","description":"Describe adjacent point triggers and their contributing positions.","command":"python -m pytest -q tests/test_analysis.py::test_anomaly_stream_detects_consecutive_run"},{"id":"isolated-events","description":"Keep separated anomalies from becoming a consecutive-run pattern.","command":"python -m pytest -q tests/test_analysis.py::test_isolated_anomalies_do_not_form_second_order_pattern"},{"id":"positional-episodes","description":"Use integer intervals for position-only episode evidence.","command":"python -m pytest -q tests/test_analysis.py::test_positional_episode_uses_integer_interval"}],"manualChecks":["Inspect observations and triggering_samples for examples/spike.json and distinguish episode grouping from pattern findings."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","criteria":[{"criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"covered","evidence":["src/anomalyzer/recipes.py:344","src/anomalyzer/recipes.py:388","tests/test_analysis.py:441"],"notes":"The injected spike retains its observed value, expectation, standardized residual, and point trigger."},{"criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"covered","evidence":["src/anomalyzer/baseline.py:31","src/anomalyzer/recipes.py:174","tests/test_analysis.py:441","tests/test_trend.py:36"],"notes":"Assertions establish repeatability and unchanged earlier evidence under later evaluation changes with fixed seasonal settings. Automatic positional season discovery has the separately documented streaming limitation."},{"criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"covered","evidence":["src/anomalyzer/evidence.py:95","tests/test_analysis.py:274"],"notes":"The test checks one three-sample run, exact contributing positions, direction, and first detection."},{"criterion":"Separated point departures remain separate episodes, and positional episode intervals use integer coordinates.","status":"covered","evidence":["src/anomalyzer/evidence.py:12","src/anomalyzer/evidence.py:23","tests/test_analysis.py:263","tests/test_analysis.py:369"],"notes":"Separated spikes produce two episodes without a consecutive-run pattern; positional evidence produces interval [9, 9]."}],"ignore_findings":[],"done":true,"input_fingerprint":"062833cc1bd19d5734b0f3c63d1e81666ffb568ead3399e359d62db2c0c0677d","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Find spikes and drops"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","label":"Know when the expected bas...","area":"analysis","scopes":["analysis","analysis/evidence"],"scope":"analysis/evidence","scopeLabel":"analysis/evidence","status":"implemented","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","acceptance":["Repeating seasonal structure with stable calibration can yield supported readiness.","Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","Insufficient calibration support yields not_assessed, including when the run otherwise completes.","Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment."],"guidance":["Readiness does not prove real-world effectiveness and is not an automatic action-policy gate."],"path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/readiness.py","src/anomalyzer/recipes.py","tests/test_readiness.py"],"automatedChecks":[{"id":"supported-reference","description":"Support repeated seasonality and preserve readiness under later changes.","command":"python -m pytest -q tests/test_readiness.py::test_repeating_season_and_trend_support_reference"},{"id":"weak-season","description":"Warn on unstable seasonality even when no findings occur.","command":"python -m pytest -q tests/test_readiness.py::test_unstable_declared_season_warns_without_calling_series_inapplicable tests/test_readiness.py::test_documented_random_example_has_no_findings_but_warns_on_reference"},{"id":"calibration-drift","description":"Warn when residual calibration drifts despite a repeated training season.","command":"python -m pytest -q tests/test_readiness.py::test_calibration_drift_warns_even_when_training_season_repeats"},{"id":"not-assessed","description":"Distinguish uncalibrated history and too few calibration samples.","command":"python -m pytest -q tests/test_readiness.py::test_readiness_is_not_assessed_without_calibrated_evaluation tests/test_readiness.py::test_short_calibration_is_not_assessed_even_when_analysis_completes"}],"manualChecks":["Read readiness reasons alongside findings; a lack of findings alone does not establish a trustworthy reference."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","criteria":[{"criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"covered","evidence":["src/anomalyzer/readiness.py:128","tests/test_readiness.py:8"],"notes":"Assertions establish repeated seasonality, strong calibration improvement, and supported readiness."},{"criterion":"Unstable declared seasonality or a drifting calibration center yields caution without making an otherwise completed run fail.","status":"covered","evidence":["src/anomalyzer/readiness.py:113","src/anomalyzer/readiness.py:122","tests/test_readiness.py:28","tests/test_readiness.py:41"],"notes":"Completed analyses retain caution reasons for weak seasonal repetition or calibration-center drift."},{"criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"covered","evidence":["src/anomalyzer/readiness.py:77","tests/test_readiness.py:56","tests/test_readiness.py:63"],"notes":"Both uncalibrated history and a completed run with three calibration samples return not_assessed."},{"criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"covered","evidence":["src/anomalyzer/recipes.py:327","tests/test_readiness.py:8"],"notes":"Readiness reads training/calibration inputs once; tests compare identical segment and aggregate assessments after a later shift."}],"ignore_findings":[],"done":true,"input_fingerprint":"d01367627e058ea39c78ee0e3616061a2e0725d06e8acd0b84b46a9655431e57","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Assess baseline quality"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","label":"Follow evolving growth whi...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","acceptance":["Each rolling expectation uses only observations before the sample being scored.","Persistent growth changes can be followed more closely than by the frozen default recipe.","An isolated extreme value is held out of future model references rather than creating a seasonal echo.","Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected."],"guidance":["A persistent third extreme miss in one direction may be admitted to the rolling reference; this is not a manual regime reset."],"path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","dependencies":["analysis/evidence/calibrate-departures"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/baseline.py","src/anomalyzer/recipes.py","src/anomalyzer/contracts.py","tests/test_analysis.py"],"automatedChecks":[{"id":"causal-window","description":"Keep a known rolling expectation invariant when future values are appended.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_expectation_is_one_step_and_prefix_invariant"},{"id":"changing-growth","description":"Follow a persistent growth change and report model admissions.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_tracks_a_persistent_growth_change"},{"id":"isolated-extreme","description":"Exclude an isolated extreme from later reference estimates.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_recipe_contains_an_isolated_extreme"},{"id":"bounded-settings","description":"Reject conflicting or unbounded adaptive settings.","command":"python -m pytest -q tests/test_analysis.py::test_adaptive_settings_reject_conflicting_or_unbounded_trend"}],"manualChecks":["Inspect the adaptive recipe's rolling-window parameters and frozen calibration before interpreting its final expectation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","criteria":[{"criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"covered","evidence":["src/anomalyzer/baseline.py:61","tests/test_analysis.py:48"],"notes":"All rolling-reference positions precede the scored index. The independent expectation remains unchanged after future values are appended."},{"criterion":"Persistent growth changes can be followed more closely than by the frozen default recipe.","status":"covered","evidence":["src/anomalyzer/recipes.py:145","src/anomalyzer/recipes.py:370","tests/test_analysis.py:58"],"notes":"The test requires adaptive held-out MAE below half the frozen recipe's MAE and a close final expectation."},{"criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","tests/test_analysis.py:81"],"notes":"The isolated spike is excluded from model references; its next seasonal expectation closely matches the control. Persistent third misses are admitted as guidance explicitly permits."},{"criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"covered","evidence":["src/anomalyzer/contracts.py:41","tests/test_analysis.py:104"],"notes":"Both incompatible explicit trend selection and excessive slope lookback raise specific validation errors."}],"ignore_findings":[],"done":true,"input_fingerprint":"984422f1fd3ab3d22cf7b4ba9d0bd06629fc228a8e53d7ad44439cd8d35ec0db","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Follow evolving growth"},"impact":0,"gaps":0,"r":38,"x":170,"y":406},{"id":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","label":"Account for daily and week...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"planned","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","acceptance":["Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","Insufficient support for a requested cycle produces an explicit limitation or declared fallback rather than a silently fitted component.","Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle."],"guidance":["Begin with a bounded daily-and-weekly use case; overlapping cycles share information and cannot be treated as independent evidence.","Proposed automated checks cover combined ordinary cycles, future isolation, unsupported cycle history, and extreme-value influence across cycles.","Manual review should inspect a weekday morning spike and verify the baseline explains ordinary calendar variation without explaining away the spike."],"path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Represent more than one repeating cycle"},"impact":0,"gaps":3,"r":47,"x":170,"y":406},{"id":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","label":"Limit how extreme values d...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","acceptance":["Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","Training and calibration diagnostics identify excluded extreme observations.","An extreme evaluated sample remains evidence while its expected value is used in future references.","Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior."],"guidance":["Protection keeps the observed evidence; it does not delete observations or certify that a departure was an incident."],"path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","dependencies":["analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/robust.py","src/anomalyzer/baseline.py","src/anomalyzer/trend.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_trend.py","tests/test_cli.py"],"automatedChecks":[{"id":"robust-reference","description":"Limit an extreme weekly reference and use earlier references at lag one.","command":"python -m pytest -q tests/test_analysis.py::test_robust_weekly_reference_limits_one_high_week tests/test_analysis.py::test_robust_lag_one_uses_prior_seasonal_values"},{"id":"training-calibration","description":"Record training and calibration exclusions and retain a useful growth fit.","command":"python -m pytest -q tests/test_trend.py::test_robust_training_fit_and_calibration_exclusions"},{"id":"include-choice","description":"Compare robust and include modes through the CLI and show their differing triggers.","command":"python -m pytest -q tests/test_cli.py::test_outlier_handling_cli"}],"manualChecks":["Inspect excluded_from_model, reference_action, and outlier-handling diagnostics alongside the original observed value."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","criteria":[{"criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"covered","evidence":["src/anomalyzer/baseline.py:39","tests/test_analysis.py:25"],"notes":"Assertions check a independently computed weighted reference and invariance when an already-trimmed high value becomes more extreme."},{"criterion":"Training and calibration diagnostics identify excluded extreme observations.","status":"covered","evidence":["src/anomalyzer/trend.py:178","src/anomalyzer/recipes.py:287","tests/test_trend.py:65"],"notes":"Tests identify exact training/calibration exclusions and compare their influence on trend selection and residual scale."},{"criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"covered","evidence":["src/anomalyzer/recipes.py:378","src/anomalyzer/recipes.py:392","tests/test_analysis.py:263","tests/test_trend.py:11"],"notes":"The observed departure and trigger survive; only reference_values receives the prior expectation, preventing the next seasonal echo."},{"criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"covered","evidence":["src/anomalyzer/recipes.py:151","src/anomalyzer/recipes.py:475","tests/test_cli.py:95"],"notes":"CLI assertions distinguish robust triggers {65} from include-mode triggers {65, 72}."}],"ignore_findings":[],"done":true,"input_fingerprint":"183b8696fdc9786878b3e54c4a9e5f68180b47e157528bef953274a3487d49a7","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Protect later comparisons"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","label":"Choose the repeating cycle...","area":"analysis","scopes":["analysis","analysis/expectations"],"scope":"analysis/expectations","scopeLabel":"analysis/expectations","status":"implemented","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","acceptance":["An explicit season length is honored and reported as explicit.","A repeating positional series can infer a seasonal lag and report the selection window.","A positional series without a sufficiently stable candidate falls back to lag one and records why.","The reserved calibration and evaluation tail does not determine positional season selection."],"guidance":["Candidate discovery uses the input prefix before the reserved tail and confirms the candidate in an early window. It is not fully prefix-invariant streaming discovery.","Supply season_length explicitly when stable streaming season selection is required."],"path":".capabilities/analysis/expectations/select-season-length.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/seasonality.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"declared-season","description":"Honor a separate explicit season setting.","command":"python -m pytest -q tests/test_analysis.py::test_bare_values_with_separate_settings"},{"id":"inference-fallback","description":"Infer a repeating lag or record lag-one fallback when no season is supported.","command":"python -m pytest -q tests/test_analysis.py::test_positional_season_inference_and_fallback"},{"id":"reserved-tail","description":"Keep selection unchanged when only the reserved tail changes.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_season_inference_excludes_reserved_tail"},{"id":"cli-inference","description":"Expose the inferred lag and its source for bare JSON values in the CLI.","command":"python -m pytest -q tests/test_cli.py::test_bare_json_values_cli"}],"manualChecks":["Review selection_window and ranked candidates for a positional series before treating its inferred season as an ongoing monitoring assumption."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","criteria":[{"criterion":"An explicit season length is honored and reported as explicit.","status":"covered","evidence":["src/anomalyzer/recipes.py:55","tests/test_analysis.py:234"],"notes":"Explicit field presence bypasses inference, and the test checks the reported source."},{"criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"covered","evidence":["src/anomalyzer/seasonality.py:96","tests/test_analysis.py:247"],"notes":"The repeating seven-position series selects lag 7 and reports confirmation window [0, 27]."},{"criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"covered","evidence":["src/anomalyzer/seasonality.py:79","src/anomalyzer/seasonality.py:84","tests/test_analysis.py:247"],"notes":"Candidate-strength and early-confirmation failures return lag 1 with a reason; the quadratic example verifies fallback."},{"criterion":"The reserved calibration and evaluation tail does not determine positional season selection.","status":"covered","evidence":["src/anomalyzer/seasonality.py:69","tests/test_capability_behaviors.py:147"],"notes":"Discovery excludes calibration_size plus one reserved sample. Changing that tail preserves the complete inference diagnostics; arbitrary appended-prefix invariance is explicitly outside the guidance."}],"ignore_findings":[],"done":true,"input_fingerprint":"3317b1727dc3cabfcfc9b87808d9249d0d59f111b8fd5f8a1fad653b21343980","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","name":"Trend and seasonal analysis","label":"Trend and seasonal analysis / Refine what ordinary behavior means / Choose a repeating cycle"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","label":"Interpret rate changes wit...","area":"analysis","scopes":["analysis","analysis/rates"],"scope":"analysis/rates","scopeLabel":"analysis/rates","status":"planned","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","acceptance":["Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","Zero denominators remain unavailable, and low support is explicitly marked; intervals remain within the valid zero-to-one rate domain.","Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations."],"guidance":["Independence and stable exposure are modeling assumptions; clustered events, overdispersion, or changing populations require explicit limitations.","Proposed automated checks cover zero denominators, tiny counts, zero-or-all-success boundaries, and equal rates at different volumes.","Manual review should compare sparse and high-volume conversion changes and confirm the explanation discloses assumptions without inventing incident probability."],"path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Judge rates using their exposure"},"impact":0,"gaps":3,"r":47,"x":194,"y":524},{"id":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","label":"See changes before the day...","area":"analysis","scopes":["analysis","analysis/resolution"],"scope":"analysis/resolution","scopeLabel":"analysis/resolution","status":"implemented","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","acceptance":["Callers can select sum, mean, or last aggregation for completed and partial periods.","Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","An explicitly unfinished source sample is excluded from both analysis views.","A source cadence that does not evenly divide 24 hours is rejected."],"guidance":["Periods are fixed 24-hour windows, not daylight-saving-aware local calendar days.","Native and daily views share source observations and are correlated.","Legacy schema 1.0 requires complete input and does not support resets. Schema 1.1 labels sparse native/completed-day views and preserves reset boundaries; daily sums require declared additive volume and fully observed children."],"path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","dependencies":["data/quality/exclude-incomplete-periods","analysis/expectations/model-seasonal-level-and-growth"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","tests/test_sparse.py","src/anomalyzer/recipes.py","tests/test_analysis.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"selected-aggregation","description":"Check sum, mean, and last for completed days and trailing partial aggregates.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_multi_resolution_uses_selected_aggregation"},{"id":"two-views","description":"Return completed days, native evidence, and a linked trailing partial-day marker.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_current_period_and_keeps_two_views"},{"id":"unfinished-source","description":"Exclude a marked incomplete source sample from both views.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_marks_partial_source_sample_after_full_days"},{"id":"dividing-cadence","description":"Reject subperiod durations that do not evenly divide a day.","command":"python -m pytest -q tests/test_analysis.py::test_multi_resolution_rejects_non_dividing_frequency"}],"manualChecks":["Inspect the partial aggregate's observed_aggregate and intraday_evidence_refs together with the two method namespaces."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","criteria":[{"criterion":"Callers can select sum, mean, or last aggregation for completed and partial periods.","status":"covered","evidence":["src/anomalyzer/contracts.py:18","src/anomalyzer/recipes.py:539","tests/test_capability_behaviors.py:178"],"notes":"Parameterized assertions check independently known completed and partial values for all three aggregation choices."},{"criterion":"Regular timestamped sub-day samples produce distinct native and completed-period analysis views.","status":"covered","evidence":["src/anomalyzer/recipes.py:617","src/anomalyzer/recipes.py:667","src/anomalyzer/recipes.py:680","tests/test_analysis.py:170"],"notes":"Separate datasets, configurations, and namespaces preserve the two correlated views."},{"criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"covered","evidence":["src/anomalyzer/recipes.py:639","src/anomalyzer/recipes.py:690","tests/test_analysis.py:170"],"notes":"The marker contains finalized progress, observed aggregate, native evidence references, and an explicit exclusion flag."},{"criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"covered","evidence":["src/anomalyzer/validation.py:44","src/anomalyzer/recipes.py:564","tests/test_analysis.py:200"],"notes":"Both child datasets use prepared finalized observations; tests verify counts and retention of the excluded original value."},{"criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"covered","evidence":["src/anomalyzer/recipes.py:572","tests/test_analysis.py:221"],"notes":"A five-hour cadence raises the specific dividing-cadence error."}],"ignore_findings":[],"done":true,"input_fingerprint":"cdcdea1166a4e3d2523b594bec938142135e17a6b4279732e52819a486d1b4f2","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","name":"Incomplete-period monitoring","label":"Incomplete-period monitoring / Monitor while periods are still open / Compare finalized subperiods"},"impact":0,"gaps":0,"r":38,"x":194,"y":524},{"id":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","label":"Compare days and weeks in ...","area":"data","scopes":["data","data/calendar"],"scope":"data/calendar","scopeLabel":"data/calendar","status":"planned","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","acceptance":["Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","Periods report UTC boundaries and local labels; daylight-saving days can contain 23 or 25 hours without being forced into fixed 24-hour windows.","Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation."],"guidance":["Calendar grouping needs a distinct contract from fixed elapsed cadence; it must not silently reinterpret existing frequency or aggregation settings.","Proposed automated checks cover spring transitions, fall transitions, week boundaries, and missing native slots in a calendar period.","Manual review should inspect a daylight-saving week and confirm local labels, observed coverage, and completed-versus-open comparisons are understandable."],"path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Declare local calendar boundaries"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","label":"Analyze many entities with...","area":"data","scopes":["data","data/cohorts"],"scope":"data/cohorts","scopeLabel":"data/cohorts","status":"planned","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","acceptance":["Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","Training, calibration, relationships, and cases do not cross entity boundaries unless a separate explicit aggregate comparison is requested.","The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident."],"guidance":["Batch orchestration should preserve existing per-request contracts rather than silently lifting schema limits; multiple-testing limits stay visible.","Proposed automated checks cover partition isolation, batch-versus-single equivalence, one-entity failure, and shared resource-budget exhaustion.","Manual review should inspect a mixed batch of healthy, changed, and sparse entities and confirm the summary does not equate missing support with health."],"path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Keep each entity's evidence separate"},"impact":0,"gaps":3,"r":47,"x":377.33333333333337,"y":524},{"id":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","label":"Inspect distinct events sh...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"implemented","summary":"Callers supply stable logical event IDs so distinct comparable calls at one timestamp retain separate values and evidence without inventing chronology.","intent":"Make real per-call evidence consumable while protecting references, replay and pattern interpretation from retry duplication and arbitrary tie order.","acceptance":["Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results."],"guidance":["Callers own identity authenticity, retry deduplication, compatible populations and any assertion of trustworthy sequence. Sorting an ID establishes no cause.","Cases and overlapping findings are inspectable numerical evidence; applications own incident deduplication, business severity and notifications."],"path":".capabilities/data/quality/identify-distinct-events.capability.yaml","dependencies":["data/inputs/accept-analysis-requests","data/quality/analyze-sparse-observations","analysis/detection/identify-residual-patterns"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/contracts.py","src/anomalyzer/events.py","src/anomalyzer/sparse.py","src/anomalyzer/orchestrator.py","src/anomalyzer/alignment.py","src/anomalyzer/relationships.py","src/anomalyzer/cases.py","src/anomalyzer/mcp_views.py","tests/test_events.py","docs/sparse-data.md","examples/sparse_examples.py","examples/render_readme_progression.py"],"automatedChecks":[{"id":"identity-and-retries","description":"Retain distinct tied calls and reject duplicate logical identities.","command":"python -m pytest -q tests/test_events.py -k \"identity or repeat or distinct or reordering\""},{"id":"honest-order","description":"Keep point/range evidence and require actual declared order for tied sequences.","command":"python -m pytest -q tests/test_events.py -k \"cohort or ties or trusted or reset or lag\""},{"id":"reproducible-replay","description":"Preserve prefix, as-of, addressed corrections and case revisions.","command":"python -m pytest -q tests/test_events.py -k \"prefix or asof or correction or revision or reference\""},{"id":"public-consumers","description":"Validate schemas, Python, CLI, MCP retrieval and bounded response growth.","command":"python -m pytest -q tests/test_events.py -k \"python_schema or bounded\""}],"manualChecks":["Inspect progression figures 23–24 and the API guide for exact original times, layout versus trusted order, retry responsibility and overlapping evidence limits."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for data/quality/identify-distinct-events.","criteria":[{"criterion":"Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","status":"partial","evidence":["src/anomalyzer/contracts.py:81","src/anomalyzer/contracts.py:118","src/anomalyzer/contracts.py:192","src/anomalyzer/contracts.py:193","src/anomalyzer/contracts.py:241","src/anomalyzer/events.py:1","src/anomalyzer/events.py:14","src/anomalyzer/events.py:22","src/anomalyzer/events.py:23","src/anomalyzer/events.py:26","src/anomalyzer/sparse.py:4","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:46","src/anomalyzer/sparse.py:60","src/anomalyzer/sparse.py:62","src/anomalyzer/orchestrator.py:21","src/anomalyzer/orchestrator.py:84","src/anomalyzer/orchestrator.py:90","src/anomalyzer/orchestrator.py:94","src/anomalyzer/orchestrator.py:100","src/anomalyzer/alignment.py:14","src/anomalyzer/alignment.py:33","src/anomalyzer/alignment.py:46","src/anomalyzer/alignment.py:47","src/anomalyzer/alignment.py:48","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:35","src/anomalyzer/cases.py:16","src/anomalyzer/cases.py:18","src/anomalyzer/cases.py:20","src/anomalyzer/cases.py:22","src/anomalyzer/cases.py:24","src/anomalyzer/mcp_views.py:68","src/anomalyzer/mcp_views.py:71","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:77","src/anomalyzer/mcp_views.py:89","tests/test_events.py:1","tests/test_events.py:22","tests/test_events.py:24","tests/test_events.py:25","tests/test_events.py:41","docs/sparse-data.md:50","docs/sparse-data.md:73","docs/sparse-data.md:74","docs/sparse-data.md:77","docs/sparse-data.md:83","examples/sparse_examples.py:36","examples/sparse_examples.py:57","examples/sparse_examples.py:65","examples/sparse_examples.py:66","examples/sparse_examples.py:69","examples/render_readme_progression.py:141","examples/render_readme_progression.py:301","examples/render_readme_progression.py:303","examples/render_readme_progression.py:373","examples/render_readme_progression.py:413"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","status":"partial","evidence":["src/anomalyzer/contracts.py:54","src/anomalyzer/contracts.py:145","src/anomalyzer/contracts.py:246","src/anomalyzer/contracts.py:255","src/anomalyzer/contracts.py:256","src/anomalyzer/events.py:26","src/anomalyzer/events.py:27","src/anomalyzer/events.py:28","src/anomalyzer/events.py:37","src/anomalyzer/events.py:46","src/anomalyzer/sparse.py:1","src/anomalyzer/sparse.py:23","src/anomalyzer/sparse.py:49","src/anomalyzer/sparse.py:54","src/anomalyzer/sparse.py:60","src/anomalyzer/orchestrator.py:11","src/anomalyzer/orchestrator.py:39","src/anomalyzer/orchestrator.py:130","src/anomalyzer/orchestrator.py:144","src/anomalyzer/orchestrator.py:170","src/anomalyzer/alignment.py:62","src/anomalyzer/relationships.py:38","src/anomalyzer/relationships.py:44","src/anomalyzer/relationships.py:164","src/anomalyzer/relationships.py:166","src/anomalyzer/relationships.py:168","src/anomalyzer/cases.py:11","src/anomalyzer/cases.py:25","src/anomalyzer/cases.py:31","src/anomalyzer/cases.py:44","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:188","src/anomalyzer/mcp_views.py:304","src/anomalyzer/mcp_views.py:322","src/anomalyzer/mcp_views.py:326","tests/test_events.py:18","tests/test_events.py:23","tests/test_events.py:30","tests/test_events.py:47","tests/test_events.py:56","docs/sparse-data.md:31","docs/sparse-data.md:88","docs/sparse-data.md:93","docs/sparse-data.md:106","docs/sparse-data.md:111","examples/sparse_examples.py:65","examples/sparse_examples.py:67","examples/sparse_examples.py:74","examples/sparse_examples.py:81","examples/sparse_examples.py:83","examples/render_readme_progression.py:346","examples/render_readme_progression.py:349","examples/render_readme_progression.py:352","examples/render_readme_progression.py:477","examples/render_readme_progression.py:542"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","status":"partial","evidence":["src/anomalyzer/contracts.py:246","src/anomalyzer/contracts.py:255","src/anomalyzer/contracts.py:256","src/anomalyzer/contracts.py:268","src/anomalyzer/contracts.py:296","src/anomalyzer/events.py:1","src/anomalyzer/events.py:50","src/anomalyzer/events.py:71","src/anomalyzer/events.py:80","src/anomalyzer/events.py:84","src/anomalyzer/sparse.py:46","src/anomalyzer/sparse.py:62","src/anomalyzer/sparse.py:93","src/anomalyzer/sparse.py:170","src/anomalyzer/sparse.py:182","src/anomalyzer/orchestrator.py:24","src/anomalyzer/orchestrator.py:142","src/anomalyzer/orchestrator.py:193","src/anomalyzer/orchestrator.py:209","src/anomalyzer/orchestrator.py:213","src/anomalyzer/alignment.py:64","src/anomalyzer/alignment.py:66","src/anomalyzer/alignment.py:68","src/anomalyzer/relationships.py:38","src/anomalyzer/relationships.py:198","src/anomalyzer/relationships.py:205","src/anomalyzer/cases.py:59","src/anomalyzer/cases.py:62","src/anomalyzer/cases.py:115","src/anomalyzer/cases.py:118","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:152","src/anomalyzer/mcp_views.py:380","src/anomalyzer/mcp_views.py:385","tests/test_events.py:29","tests/test_events.py:30","tests/test_events.py:80","tests/test_events.py:81","tests/test_events.py:121","docs/sparse-data.md:118","docs/sparse-data.md:136","docs/sparse-data.md:137","docs/sparse-data.md:138","docs/sparse-data.md:139","examples/sparse_examples.py:87","examples/sparse_examples.py:88","examples/sparse_examples.py:89","examples/render_readme_progression.py:918","examples/render_readme_progression.py:921","examples/render_readme_progression.py:1024","examples/render_readme_progression.py:1037","examples/render_readme_progression.py:1038"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results.","status":"partial","evidence":["src/anomalyzer/contracts.py:308","src/anomalyzer/contracts.py:315","src/anomalyzer/contracts.py:319","src/anomalyzer/contracts.py:596","src/anomalyzer/contracts.py:597","src/anomalyzer/events.py:1","src/anomalyzer/events.py:14","src/anomalyzer/events.py:100","src/anomalyzer/events.py:102","src/anomalyzer/events.py:165","src/anomalyzer/sparse.py:182","src/anomalyzer/sparse.py:194","src/anomalyzer/sparse.py:195","src/anomalyzer/sparse.py:202","src/anomalyzer/sparse.py:333","src/anomalyzer/orchestrator.py:122","src/anomalyzer/orchestrator.py:143","src/anomalyzer/orchestrator.py:155","src/anomalyzer/orchestrator.py:162","src/anomalyzer/orchestrator.py:164","src/anomalyzer/alignment.py:63","src/anomalyzer/alignment.py:66","src/anomalyzer/relationships.py:35","src/anomalyzer/relationships.py:198","src/anomalyzer/relationships.py:203","src/anomalyzer/relationships.py:205","src/anomalyzer/cases.py:29","src/anomalyzer/cases.py:30","src/anomalyzer/cases.py:33","src/anomalyzer/cases.py:37","src/anomalyzer/cases.py:62","src/anomalyzer/mcp_views.py:12","src/anomalyzer/mcp_views.py:14","src/anomalyzer/mcp_views.py:76","src/anomalyzer/mcp_views.py:77","src/anomalyzer/mcp_views.py:88","tests/test_events.py:1","tests/test_events.py:12","tests/test_events.py:36","tests/test_events.py:49","tests/test_events.py:51","docs/sparse-data.md:37","docs/sparse-data.md:97","docs/sparse-data.md:100","docs/sparse-data.md:111","docs/sparse-data.md:115","examples/sparse_examples.py:39","examples/sparse_examples.py:72","examples/sparse_examples.py:73","examples/sparse_examples.py:77","examples/sparse_examples.py:80","examples/render_readme_progression.py:3","examples/render_readme_progression.py:781","examples/render_readme_progression.py:782","examples/render_readme_progression.py:787","examples/render_readme_progression.py:798"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","partial: Without a trusted sequence, whole timestamp cohorts share prior-time point references and ties cannot establish change, Nelson or CUSUM continuity.","partial: Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","partial: Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","impact":0,"gaps":0,"r":38,"x":377.33333333333337,"y":524},{"id":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","label":"Know when missing updates ...","area":"data","scopes":["data","data/quality"],"scope":"data/quality","scopeLabel":"data/quality","status":"planned","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","acceptance":["Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","Results distinguish current, late, missing, and not-assessed freshness and expose the last observed update and the deadline used.","A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision."],"guidance":["Freshness depends on declared collection expectations; it does not prove source completeness and does not start a scheduler or send a notification.","Proposed automated checks cover exact deadlines, allowed lateness, absent source updates, and stale-source propagation to relationships.","Manual review should compare a stale feed with a true metric drop and confirm the report distinguishes collection uncertainty from numerical departure."],"path":".capabilities/data/quality/report-source-freshness.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","name":"Future - Calendar and data readiness","label":"Future - Calendar and data readiness / Monitor the periods my business uses / Check whether the sources are current"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","label":"See how late or corrected ...","area":"data","scopes":["data","data/revisions"],"scope":"data/revisions","scopeLabel":"data/revisions","status":"planned","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","acceptance":["A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","Results preserve prior and new evidence versions and identify findings that were added, changed, or withdrawn with traceable source lineage.","Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results."],"guidance":["Schema-1.1 snapshot revision selection and as-of availability are implemented in data/quality/analyze-sparse-observations; durable prior/new evidence diffing and automatic reconciliation in this capability remain planned.","A corrected event-time result does not replace the record of what was known earlier; callers control retention and any notification about a revision.","Proposed automated checks cover late completion, corrected calibration data, downstream relationship changes, and conflicting source versions.","Manual review should inspect a drop withdrawn after ingestion completes and confirm both the original limitation and the revision explanation are retained."],"path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Reconcile revisions without losing history"},"impact":0,"gaps":3,"r":47,"x":353.33333333333337,"y":642},{"id":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","label":"Choose settings using held...","area":"evaluation","scopes":["evaluation","evaluation/comparisons"],"scope":"evaluation/comparisons","scopeLabel":"evaluation/comparisons","status":"planned","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","acceptance":["Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","The output preserves every candidate's settings and evaluation provenance; adopting a configuration requires an explicit caller decision."],"guidance":["Bounded comparison is not unrestricted parameter search; held-out results describe the supplied history and do not guarantee future effectiveness.","Proposed automated checks cover split isolation, comparable replay inputs, failed candidate retention, and unchanged active settings after comparison.","Manual review should compare an earlier but noisier configuration with a quieter one and confirm the tradeoff is visible without an unsupported winner."],"path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Review the tradeoff before adopting settings"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","label":"Measure detection timing a...","area":"evaluation","scopes":["evaluation","evaluation/outcomes"],"scope":"evaluation/outcomes","scopeLabel":"evaluation/outcomes","status":"planned","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","acceptance":["Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","Timing reports distinguish the incident's supplied start from the first available candidate, calibrated detection, and policy-eligible evidence.","Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown."],"guidance":["Detection timing does not measure response effectiveness or establish that an unmatched finding was false; incident labels are caller judgments.","Proposed automated checks cover boundary matching, ambiguous overlaps, incomplete label coverage, and separate candidate versus eligibility timing.","Manual review should inspect a matched and an unmatched episode and confirm the reported timing and accuracy limits agree with the supplied label coverage."],"path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Compare detection with reviewed outcomes"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","label":"Replay history as it would...","area":"evaluation","scopes":["evaluation","evaluation/replay"],"scope":"evaluation/replay","scopeLabel":"evaluation/replay","status":"planned","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","acceptance":["Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","Repeating a replay preserves numerical evidence and first-detection timing apart from generated IDs and runtime; stopped replays retain completed steps."],"guidance":["Event-time replay does not reconstruct arrival-time availability unless the caller supplies versioned arrival history.","Proposed automated checks cover future-tail isolation, short-history cutoffs, deterministic repeated replay, and interrupted replay retention.","Manual review should inspect one change from its first available observation through calibration and confirm the explanation uses only then-known evidence."],"path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","name":"Future - Historical evaluation","label":"Future - Historical evaluation / Evaluate a monitor before relying on it / Reconstruct the evidence available at the time"},"impact":0,"gaps":3,"r":47,"x":536.6666666666667,"y":642},{"id":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","label":"Update evidence when the n...","area":"execution","scopes":["execution","execution/incremental"],"scope":"execution/incremental","scopeLabel":"execution/incremental","status":"planned","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","acceptance":["Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","Supported explicitly configured recipes produce the same numerical evidence as full batch analysis of the combined history, including cross-chunk patterns.","Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations."],"guidance":["Start with explicit season lengths; input-prefix season discovery is not currently guaranteed to be invariant when the history grows.","Proposed automated checks cover chunk-versus-batch equivalence, cross-chunk CUSUM, idempotent retry, and failed-append state preservation.","Manual review should follow a sustained change across several append batches and confirm detection timing and provisional evidence remain understandable."],"path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Analyze the next finalized samples"},"impact":0,"gaps":3,"r":47,"x":720,"y":642},{"id":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","label":"Keep useful evidence when ...","area":"execution","scopes":["execution","execution/resilience"],"scope":"execution/resilience","scopeLabel":"execution/resilience","status":"implemented","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","acceptance":["A method failure preserves evidence completed before the failure and reports the error.","Runtime budgets are checked cooperatively and produce an explicit stop reason.","CLI byte limits and execution failures produce distinct error exits.","A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial."],"guidance":["Default limits are 10,000 source observations per request, 10 MB CLI input, and 60 seconds cooperative runtime.","Numerical results are deterministic for fixed declared settings; generated run IDs and runtimes vary."],"path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","dependencies":["data/inputs/accept-analysis-requests"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/budget.py","src/anomalyzer/recipes.py","src/anomalyzer/orchestrator.py","src/anomalyzer/io.py","src/anomalyzer/cli.py","tests/test_analysis.py","tests/test_cli.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"method-and-budget","description":"Retain pre-failure evidence and report a runtime-budget stop.","command":"python -m pytest -q tests/test_analysis.py::test_failure_and_budget"},{"id":"cli-limits","description":"Enforce byte limits and distinguish execution failure in CLI output and exit status.","command":"python -m pytest -q tests/test_cli.py::test_size_limit_and_execution_exit"},{"id":"partial-siblings","description":"Preserve healthy analysis when a sibling dataset fails preparation.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_partial_analysis_preserves_healthy_siblings"}],"manualChecks":["Inspect run status, stop_reason, and each target's status before relying on a result; confirm limits are presented as cooperative rather than hard process timeouts."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","criteria":[{"criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"covered","evidence":["src/anomalyzer/recipes.py:400","src/anomalyzer/recipes.py:448","tests/test_analysis.py:509"],"notes":"The failure regression preserves evidence through index 44 and reports the injected error at index 45."},{"criterion":"Runtime budgets are checked cooperatively and produce an explicit stop reason.","status":"covered","evidence":["src/anomalyzer/budget.py:18","src/anomalyzer/recipes.py:108","src/anomalyzer/recipes.py:508","tests/test_analysis.py:509"],"notes":"Cooperative target/sample checks expose runtime_budget and budget_exceeded; this is not a hard process timeout."},{"criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"covered","evidence":["src/anomalyzer/io.py:9","src/anomalyzer/cli.py:429","tests/test_cli.py:108"],"notes":"The full suite executed the byte-limit and execution-exit regression successfully; invalid input exits 2 and execution failure exits 1."},{"criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"covered","evidence":["src/anomalyzer/orchestrator.py:125","src/anomalyzer/orchestrator.py:171","src/anomalyzer/orchestrator.py:185","tests/test_capability_behaviors.py:159","tests/test_capability_behaviors.py:30"],"notes":"Healthy dataset evidence survives target-specific failure; inspected incomplete-usage reports partial with no invented healthy relationship."}],"ignore_findings":[],"done":true,"input_fingerprint":"146a9c8396fd7a2d039f75077dc0b82a0778685d20c3782aa284d5f0b6ed94be","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Inspect incomplete execution"},"impact":0,"gaps":0,"r":38,"x":720,"y":642},{"id":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","label":"Resume local analysis afte...","area":"execution","scopes":["execution","execution/state"],"scope":"execution/state","scopeLabel":"execution/state","status":"planned","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","acceptance":["A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","Restoring a supported state and continuing yields the same numerical evidence and detection positions as uninterrupted processing of the same observations.","Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled."],"guidance":["Stored observations and context can contain sensitive data; document exactly what a snapshot contains and avoid automatic persistence.","Proposed automated checks cover restart equivalence, corrupt snapshots, version or settings mismatch, and output overwrite protection.","Manual review should inspect a snapshot and resumed explanation and confirm their provenance and retained-data scope are clear to an application owner."],"path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","name":"Future - Continuous analysis","label":"Future - Continuous analysis / Keep evidence current as observations arrive / Preserve state across restarts"},"impact":0,"gaps":3,"r":47,"x":744,"y":760},{"id":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","label":"Understand findings in a r...","area":"interfaces","scopes":["interfaces","interfaces/cli"],"scope":"interfaces/cli","scopeLabel":"interfaces/cli","status":"implemented","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","acceptance":["Point-anomaly episode counts and residual-pattern findings are labeled separately.","Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","Readiness status remains visible, including when it was not assessed."],"guidance":["Readable summaries should retain the evidence's claim limits and allow drilldown into JSON."],"path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/cli.py","tests/test_cli.py","examples/agentic/conversion.json","examples/agentic/unit-cost.json"],"automatedChecks":[{"id":"episodes-patterns","description":"Distinguish shift patterns from point episodes in readable output.","command":"python -m pytest -q tests/test_cli.py::test_readable_output_distinguishes_episodes_from_patterns"},{"id":"relationship-case","description":"Explain a conversion case with entity, units, comparison, and a suggested check.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_relationship_request_through_cli"},{"id":"explicit-rule","description":"Describe the unit-cost rule boundary and unassessed readiness.","command":"python -m pytest -q tests/test_cli.py::test_schema_11_readable_case_distinguishes_explicit_rule"}],"manualChecks":["Read the conversion and unit-cost example output and confirm the descriptions state the supported numerical comparison in plain language."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","criteria":[{"criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"covered","evidence":["src/anomalyzer/cli.py:347","src/anomalyzer/cli.py:405","tests/test_cli.py:143"],"notes":"Readable tests distinguish zero-episode residual shifts from a spike episode and retain the overlap warning."},{"criterion":"Relationship cases identify entity scope, observed and expected values, declared units, and investigation questions.","status":"covered","evidence":["src/anomalyzer/cli.py:291","src/anomalyzer/cli.py:308","src/anomalyzer/cli.py:331","tests/test_cli.py:117","examples/agentic/conversion.json"],"notes":"Conversion output identifies site=demo, observed/expected orders_per_visit, the relationship, and a collection-completeness investigation question."},{"criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"covered","evidence":["src/anomalyzer/cli.py:318","tests/test_cli.py:134","examples/agentic/unit-cost.json"],"notes":"Unit-cost output names review-above-13c and its [0, 0.13] boundary as a declared rule crossing."},{"criterion":"Readiness status remains visible, including when it was not assessed.","status":"covered","evidence":["src/anomalyzer/cli.py:352","src/anomalyzer/cli.py:399","tests/test_cli.py:134"],"notes":"Readable output includes not_assessed and its supporting reasons without claiming measured alert accuracy."}],"ignore_findings":[],"done":true,"input_fingerprint":"98108868b000972fd9b3c3f082846fbd3f3b08c399ef7ed424c23245d6935fef","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","name":"Basic CLI anomaly detection","label":"Basic CLI anomaly detection / Analyze a local series / Read the findings"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","label":"Share an investigation tha...","area":"interfaces","scopes":["interfaces","interfaces/export"],"scope":"interfaces/export","scopeLabel":"interfaces/export","status":"planned","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","acceptance":["Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","Source observations and context are included only by explicit selection; omitted data and resulting limits on independent reproduction are disclosed."],"guidance":["Fingerprints identify included provenance; they do not prove source truth. A portable report must make privacy scope and reproduction limits inspectable.","Proposed automated checks cover manifest consistency, partial-run export, omitted-source disclosure, and overwrite protection.","Manual review should open a bundle outside the original session and confirm a reviewer can trace one finding and see exactly what cannot be reproduced."],"path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Preserve a portable evidence record"},"impact":0,"gaps":3,"r":47,"x":927.3333333333334,"y":760},{"id":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","label":"Install a versioned analys...","area":"interfaces","scopes":["interfaces","interfaces/python"],"scope":"interfaces/python","scopeLabel":"interfaces/python","status":"implemented","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","acceptance":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"guidance":["Guard owns scheduling, persistence, business severity, incidents and notifications.","A published package does not establish integration into UsageTap Guard."],"path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","dependencies":["interfaces/python/analyze-through-public-api","interfaces/cli/analyze-files-and-streams","interfaces/mcp/connect-read-only-local-server"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["pyproject.toml","src/anomalyzer/_version.py","scripts/release.py","scripts/publish_release.py","scripts/verify_distribution.py","scripts/verify_lambda.sh","examples/library/guard_smoke.py",".github/workflows/ci.yml",".github/workflows/release.yml","docs/releases/README.md","tests/test_release.py"],"automatedChecks":[{"id":"release-safety","description":"Reject mismatched tags, altered assets and mutable publication.","command":"python -m pytest -q tests/test_release.py"},{"id":"installed-artifacts","description":"Install wheel and sdist outside the checkout and invoke public APIs.","command":"python scripts/verify_distribution.py --python 3.12"},{"id":"release-inventory","description":"Check archive contents, version metadata and SHA-256 inventory.","command":"python scripts/release.py verify"},{"id":"linux-offline-consumer","description":"Bundle and run with networking disabled in Linux Python 3.12.","command":"bash scripts/verify_lambda.sh"}],"manualChecks":["Follow the release guide and confirm Guard retains domain severity and incident ownership; inspect readiness and evidence limits before adopting."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for interfaces/python/install-versioned-library.","criteria":[{"criterion":"Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","status":"partial","evidence":["src/anomalyzer/_version.py:1","scripts/release.py:1","scripts/release.py:20","scripts/release.py:39","scripts/release.py:48","scripts/release.py:54","scripts/publish_release.py:21","scripts/publish_release.py:34","scripts/publish_release.py:42","scripts/publish_release.py:53","scripts/verify_lambda.sh:15","scripts/verify_lambda.sh:18","scripts/verify_lambda.sh:20","scripts/verify_lambda.sh:26","scripts/verify_lambda.sh:28","examples/library/guard_smoke.py:74",".github/workflows/ci.yml:14",".github/workflows/ci.yml:19",".github/workflows/ci.yml:27",".github/workflows/ci.yml:43",".github/workflows/ci.yml:48",".github/workflows/release.yml:31",".github/workflows/release.yml:52",".github/workflows/release.yml:61","docs/releases/README.md:16","docs/releases/README.md:17","docs/releases/README.md:18","docs/releases/README.md:21","docs/releases/README.md:22","tests/test_release.py:60","tests/test_release.py:71","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","status":"partial","evidence":["scripts/verify_distribution.py:1","scripts/verify_distribution.py:48","examples/library/guard_smoke.py:32","examples/library/guard_smoke.py:49","examples/library/guard_smoke.py:65","examples/library/guard_smoke.py:69","examples/library/guard_smoke.py:76",".github/workflows/ci.yml:59",".github/workflows/release.yml:60","docs/releases/README.md:17","docs/releases/README.md:99","docs/releases/README.md:217","docs/releases/README.md:220","docs/releases/README.md:222"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:34","scripts/release.py:81","scripts/release.py:82","scripts/release.py:85","scripts/release.py:92","scripts/publish_release.py:1","scripts/publish_release.py:18","scripts/publish_release.py:19","scripts/publish_release.py:20","scripts/publish_release.py:40","scripts/verify_distribution.py:47","scripts/verify_lambda.sh:25","examples/library/guard_smoke.py:39",".github/workflows/ci.yml:44",".github/workflows/ci.yml:47",".github/workflows/ci.yml:48",".github/workflows/ci.yml:51",".github/workflows/ci.yml:53",".github/workflows/release.yml:34",".github/workflows/release.yml:36",".github/workflows/release.yml:39",".github/workflows/release.yml:40",".github/workflows/release.yml:48","docs/releases/README.md:8","docs/releases/README.md:16","docs/releases/README.md:18","docs/releases/README.md:19","docs/releases/README.md:23","tests/test_release.py:1","tests/test_release.py:21","tests/test_release.py:23","tests/test_release.py:52","tests/test_release.py:81"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","status":"partial","evidence":["pyproject.toml:26","pyproject.toml:33","scripts/release.py:81","scripts/release.py:82","scripts/publish_release.py:54","scripts/verify_distribution.py:47","scripts/verify_lambda.sh:6","scripts/verify_lambda.sh:18","scripts/verify_lambda.sh:20",".github/workflows/ci.yml:29",".github/workflows/ci.yml:32",".github/workflows/ci.yml:33",".github/workflows/ci.yml:53",".github/workflows/ci.yml:54",".github/workflows/release.yml:62","docs/releases/README.md:16","docs/releases/README.md:19","docs/releases/README.md:62","docs/releases/README.md:63","docs/releases/README.md:79"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Wheel and source distributions carry the complete Python package, license, public API, CLI and read-only MCP entry points with one authoritative version.","partial: Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","partial: Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","partial: A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","name":"Python integration","label":"Python integration / Embed analysis in my application / Install a released dependency"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","label":"Ask an agent to explain un...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","acceptance":["The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit."],"guidance":["This contract covers the supplied instructions and demonstrated analysis workflow; it does not guarantee that every agent host follows the instructions.","Synthetic examples demonstrate mechanics, not real-world detection accuracy."],"path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams","interfaces/mcp/investigate-findings-and-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/agents/openai.yaml","src/anomalyzer/cli.py","src/anomalyzer/mcp.py","tests/test_cli.py","tests/test_agentic.py","tests/test_mcp.py"],"automatedChecks":[{"id":"structured-cli","description":"Exercise the documented structured CLI workflow and preserve incomplete-history evidence.","command":"python -m pytest -q tests/test_cli.py::test_stdin_file_equivalence tests/test_cli.py::test_bare_json_values_cli"},{"id":"declared-relationships","description":"Check explicit ratio semantics and unavailable source evidence.","command":"python -m pytest -q tests/test_agentic.py::test_ratio_lineage_and_early_relationship_departure"},{"id":"mcp-drilldown","description":"Follow compact analysis to a finding and its source evidence.","command":"python -m pytest -q tests/test_mcp.py::test_agent_can_drill_into_relationship_case"}],"manualChecks":["Follow the conversion example through the CLI and MCP drilldown; compare the Skill instructions with actual evidence and check that its proposed text separates calibrated findings, early candidates, readiness, and caller-owned actions."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","criteria":[{"criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:13",".agents/skills/anomaly-detection/SKILL.md:41","tests/test_cli.py:13"],"notes":"Executable resolution and JSON commands are explicit. Manual conversion and spike CLI runs use the installed numerical engine."},{"criterion":"The Skill instructs callers to preserve missing data, select one entity and unit, and explicitly declare compatible sources for related-metric analysis.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:24",".agents/skills/anomaly-detection/SKILL.md:32","tests/test_cli.py:51","tests/test_agentic.py:211"],"notes":"Instructions preserve scope, units, missingness, declared relationships, and exact alignment; existing runtime checks establish these prerequisites."},{"criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:101",".agents/skills/anomaly-detection/SKILL.md:134",".agents/skills/anomaly-detection/agents/openai.yaml:4"],"notes":"The supplied instructions lead with plain-language changes and then qualify maturity/readiness/overlap. This establishes instructional content, not measured agent compliance."},{"criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:54",".agents/skills/anomaly-detection/SKILL.md:174","src/anomalyzer/mcp.py:169","tests/test_mcp.py:54"],"notes":"The documented three-tool workflow matches executed conversion drilldown, with session-only run IDs and explicit read-only/caller-owned delivery limits."}],"ignore_findings":[],"done":true,"input_fingerprint":"751723214aa340b1227ec347c22f130749fcdbfd49327c21b4c27ef68d7f0341","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Get an evidence-grounded explanation"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","label":"See the evidence in a char...","area":"interfaces","scopes":["interfaces","interfaces/skill"],"scope":"interfaces/skill","scopeLabel":"interfaces/skill","status":"implemented","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","acceptance":["The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes."],"guidance":["Native chart rendering and interactive figure features depend on the agent host; the checked-in fallback produces a static SVG for a single dataset.","A chart explains structured analysis; it does not replace the detector result."],"path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","dependencies":["interfaces/cli/analyze-files-and-streams"],"suggestedDependencies":[],"dependents":[],"implementationReferences":[".agents/skills/anomaly-detection/SKILL.md",".agents/skills/anomaly-detection/scripts/quick_chart.py",".agents/skills/anomaly-detection/references/visualization.md","tests/test_skill_chart.py"],"automatedChecks":[{"id":"observed-evidence","description":"Render real observed and expected data and only actual point triggers.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_preserves_real_point_evidence"},{"id":"maturity-and-resets","description":"Keep provisional evidence non-triggering and label reviewed regimes.","command":"python -m pytest -q tests/test_skill_chart.py::test_provisional_scores_are_not_charted_as_triggers tests/test_skill_chart.py::test_quick_chart_marks_reviewed_resets"},{"id":"unsupported-scope","description":"Reject a multi-dataset result rather than mixing units.","command":"python -m pytest -q tests/test_skill_chart.py::test_quick_chart_rejects_multiple_datasets"},{"id":"output-protection","description":"Require an explicit overwrite for existing charts.","command":"python -m pytest -q tests/test_skill_chart.py::test_chart_cli_requires_explicit_overwrite"}],"manualChecks":["Inspect a spike fallback chart and the detailed figure instructions; verify that axes, initialization context, trigger markers, and interpretation limits are clear, and that the Skill requests a chart only when the user asks."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json.","Skill criteria covering agent instructions are supported by the written guidance and demonstrated tool workflow; cross-host agent compliance and real-world detection accuracy are not asserted."],"intent_summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","criteria":[{"criterion":"The Skill defaults to text and requires an explicit visualization request; quick charts and publication-quality figures have separate instructions.","status":"covered","evidence":[".agents/skills/anomaly-detection/SKILL.md:65",".agents/skills/anomaly-detection/SKILL.md:189",".agents/skills/anomaly-detection/SKILL.md:226"],"notes":"The Skill explicitly requires a visual request after analysis and separates quick versus detailed rendering instructions."},{"criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:81",".agents/skills/anomaly-detection/scripts/quick_chart.py:59","tests/test_skill_chart.py:28","tests/test_skill_chart.py:43","tests/test_skill_chart.py:54"],"notes":"Real-result tests establish observed/expected lines, trigger-only markers, provisional non-triggering, units, initialization shading, and reset boundaries; the spike chart was inspected visually."},{"criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"covered","evidence":[".agents/skills/anomaly-detection/scripts/quick_chart.py:85",".agents/skills/anomaly-detection/scripts/quick_chart.py:251","tests/test_skill_chart.py:63","tests/test_skill_chart.py:68"],"notes":"Multiple resolved datasets raise a specific error. The CLI test preserves an existing chart and replaces it only with explicit overwrite."},{"criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"covered","evidence":[".agents/skills/anomaly-detection/references/visualization.md:31",".agents/skills/anomaly-detection/references/visualization.md:37",".agents/skills/anomaly-detection/references/visualization.md:45",".agents/skills/anomaly-detection/references/visualization.md:86"],"notes":"Detailed instructions require index alignment, maturity distinctions, first detection timing, visible unavailable status, and qualified causal claims; native interactive rendering remains host-dependent."}],"ignore_findings":[],"done":true,"input_fingerprint":"2780b6c3e363f89a30a0709772c0a46a9a96f1567fabb27e63934a3ca80dc982","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","name":"Anomaly detection Skill","label":"Anomaly detection Skill / Ask an agent to investigate a change / Inspect a requested chart"},"impact":0,"gaps":0,"r":38,"x":927.3333333333334,"y":760},{"id":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","label":"Find which segments contri...","area":"investigation","scopes":["investigation","investigation/cohorts"],"scope":"investigation/cohorts","scopeLabel":"investigation/cohorts","status":"planned","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","acceptance":["Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","Segment ranking retains absolute signed contributions and baseline support; an arithmetic contributor is not automatically labeled anomalous or causal."],"guidance":["Ratios and overlapping customer groups are not additive totals; separate segment expectations must not be assumed to sum to a fitted total expectation.","Proposed automated checks cover exact reconciliation, offsetting segment changes, missing segments, and zero total departure.","Manual review should inspect a regional decline with an offsetting increase and confirm the explanation preserves signed contributions and evidence limits."],"path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Locate contributors to a total change"},"impact":0,"gaps":3,"r":47,"x":1110.6666666666667,"y":760},{"id":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","label":"Keep reviewed outcomes bes...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","acceptance":["A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","Review export distinguishes analyst judgment from detector output; annotation alone does not change calibration, eligibility, or regime boundaries."],"guidance":["Caller-supplied reviewer identity is attribution, not built-in authentication; storage, access control, response execution, and incident ownership stay with callers.","Proposed automated checks cover evidence-version linking, unresolved reviews, revision retention, and unchanged numerical evidence after annotation.","Manual review should inspect a disputed case with two review revisions and confirm judgments, uncertainties, and original findings remain distinguishable."],"path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Record the reviewed outcome"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","label":"Find earlier reviewed case...","area":"investigation","scopes":["investigation","investigation/history"],"scope":"investigation/history","scopeLabel":"investigation/history","status":"planned","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","acceptance":["Search is bounded to caller-selected records and explicit compatibility filters; it does not scan unrelated local files or require a model API.","Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match."],"guidance":["Similar historical evidence does not establish the same cause or authorize repeating a past response; caller-selected records may be incomplete or biased.","Proposed automated checks cover compatibility filtering, bounded corpus access, missing comparison context, and no-match behavior.","Manual review should compare superficially similar cases with different outcomes and confirm precedent is presented as investigation context."],"path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","name":"Future - Investigation memory","label":"Future - Investigation memory / Reuse what an investigation teaches us / Inspect comparable reviewed history"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","label":"Change action eligibility ...","area":"investigation","scopes":["investigation","investigation/policy"],"scope":"investigation/policy","scopeLabel":"investigation/policy","status":"implemented","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","acceptance":["Default policy leaves action and notification eligibility disabled.","Explicit-rule and calibrated-departure eligibility are controlled independently.","Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged."],"guidance":["Eligibility does not send notifications, execute actions, check readiness, or confer authority to act.","Monitoring schedules, delivery, reviewed incident state, and deduplication belong to the caller."],"path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","dependencies":["investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/agent.py","src/anomalyzer/cases.py","src/anomalyzer/contracts.py","tests/test_agentic.py","tests/test_capability_behaviors.py"],"automatedChecks":[{"id":"independent-notifications","description":"Replay notification and action choices without changing numerical evidence or the original result.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_separates_action_from_notification"},{"id":"calibrated-choice","description":"Enable calibrated departures only with the corresponding policy choice.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_replay_enables_calibrated_departures_only_when_requested"},{"id":"nontriggering-candidates","description":"Keep descriptive candidates ineligible even under a permissive policy.","command":"python -m pytest -q tests/test_capability_behaviors.py::test_policy_keeps_early_candidates_ineligible"}],"manualChecks":["Inspect the unit-cost example under default and allow_explicit_rules policies and compare eligibility with the unchanged numerical violation."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","criteria":[{"criterion":"Default policy leaves action and notification eligibility disabled.","status":"covered","evidence":["src/anomalyzer/contracts.py:329","tests/test_capability_behaviors.py:74"],"notes":"Default flags disable both eligibility types, including after replay resets policy."},{"criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"covered","evidence":["src/anomalyzer/agent.py:47","tests/test_capability_behaviors.py:123"],"notes":"Explicit-only permission does not enable a calibrated departure; the matching calibrated choice does."},{"criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"covered","evidence":["src/anomalyzer/agent.py:53","tests/test_capability_behaviors.py:101","tests/test_capability_behaviors.py:134"],"notes":"Notification-only replay enables a met rule while keeping action disabled; permissive settings still cannot enable candidates."},{"criterion":"Replaying policy preserves numerical evidence and stable case IDs, records changed case revisions, and leaves the original result unchanged.","status":"covered","evidence":["src/anomalyzer/agent.py:42","src/anomalyzer/cases.py:57","src/anomalyzer/cases.py:105","tests/test_capability_behaviors.py:101","tests/test_agentic.py:49"],"notes":"Assertions compare unchanged original serialization, methods, evidence references, stable IDs, and eligibility-dependent revision changes."}],"ignore_findings":[],"done":true,"input_fingerprint":"ee67bcc1ec1380b8d84a0556d838d648ae6cef3f4f9b9e34e1b71e90eb00aeae","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","name":"Evidence and review policy","label":"Evidence and review policy / Decide how much to rely on a finding / Review action eligibility"},"impact":0,"gaps":0,"r":38,"x":1086.6666666666667,"y":878},{"id":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","label":"Review a possible new base...","area":"investigation","scopes":["investigation","investigation/regimes"],"scope":"investigation/regimes","scopeLabel":"investigation/regimes","status":"planned","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","acceptance":["A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","Producing a proposal leaves the active regime and original findings unchanged.","A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect."],"guidance":["A proposed boundary is not a confirmed cause or a retrospectively available detection; adaptive reference updates remain a separate behavior.","Proposed automated checks cover unchanged active results, short new regimes, rejected proposals, and accepted explicit-reset equivalence.","Manual review should inspect a persistent outage-like drop and a known launch and confirm neither becomes ordinary behavior merely because it persists."],"path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","name":"Future - Business-aware expectations","label":"Future - Business-aware expectations / Make expectations fit the process / Review a proposed regime transition"},"impact":0,"gaps":3,"r":47,"x":1086.6666666666667,"y":878},{"id":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","label":"Distinguish changing perfo...","area":"relationships","scopes":["relationships","relationships/cohorts"],"scope":"relationships/cohorts","scopeLabel":"relationships/cohorts","status":"planned","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","acceptance":["Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","The report preserves aggregate and cohort rates, population weights, and the declared decomposition formula and reference-period convention.","Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes."],"guidance":["Decomposition is descriptive accounting; it does not estimate an intervention effect or establish that the comparison populations are causally equivalent.","Proposed automated checks cover stable rates with changing mix, stable mix with changing rates, offsetting cohort rates, and new or empty cohorts.","Manual review should inspect a Simpson's-paradox example and confirm the aggregate movement and cohort evidence can both be understood correctly."],"path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","dependencies":[],"suggestedDependencies":[],"dependents":[],"implementationReferences":[],"automatedChecks":[],"manualChecks":[],"verificationGaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}],"reviewFindings":[],"reviewStale":false,"health":"gap","healthLabel":"Needs action","coverage":"uncovered","storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","name":"Future - Cohort investigation","label":"Future - Cohort investigation / Investigate changes across populations / Separate performance changes from population mix"},"impact":0,"gaps":3,"r":47,"x":1270,"y":878},{"id":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","label":"Check whether a metric fol...","area":"relationships","scopes":["relationships","relationships/comparisons"],"scope":"relationships/comparisons","scopeLabel":"relationships/comparisons","status":"implemented","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","acceptance":["A supplied lag and coefficient use the predictor at the declared earlier sample position.","Derived values retain response and predictor source indexes and source cutoffs in lineage.","When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","Samples lacking the earlier predictor or awaiting coefficient training remain unavailable."],"guidance":["Lag is measured in samples and is never searched automatically.","Exactly one of coefficient or training_size is required."],"path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","dependencies":["relationships/quality/require-compatible-sources"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/relationships.py","src/anomalyzer/orchestrator.py","tests/test_agentic.py","tests/test_examples.py","examples/agentic/lagged-response.json"],"automatedChecks":[{"id":"supplied-response","description":"Use the declared lag and coefficient with correct predictor indexes.","command":"python -m pytest -q tests/test_agentic.py::test_lagged_response_uses_supplied_coefficient_and_prior_sample"},{"id":"frozen-coefficient","description":"Keep fitted coefficient and earlier lineage invariant when future samples arrive.","command":"python -m pytest -q tests/test_agentic.py::test_estimated_lag_coefficient_is_frozen_and_prefix_invariant"},{"id":"delayed-example","description":"Run the checked-in lagged-response demonstration against expected results.","command":"python -m pytest -q tests/test_examples.py::test_agentic_demo_checks_all_cases"},{"id":"pending-training","description":"Retain unavailable lineage until the full declared training prefix arrives, then freeze its coefficient.","command":"python -m pytest -q tests/test_agentic.py::test_lag_fit_waits_for_complete_training_prefix"}],"manualChecks":["Inspect the lagged-response example's leading unavailable values, coefficient parameters, and first usable derived sample."],"verificationGaps":[],"review":{"depth":"verified","source":"coding-agent","gaps":[],"evidence":["Primary Codex agent re-inspected declared implementation references and criterion evidence in the current session after the outcome-title and deliverable-lane refactor.",".venv/Scripts/python.exe -m pytest -q: 154 passed in 13.98s; .capabilities/reviews/deliverable-lanes/verification.json.","Executed and inspected CLI examples, time preparation, policy replay, open-day monitoring, and MCP conversion evidence drilldown; .capabilities/reviews/deliverable-lanes/manual-checks.json."],"intent_summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","criteria":[{"criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"covered","evidence":["src/anomalyzer/relationships.py:115","tests/test_agentic.py:254"],"notes":"Supplied lag/coefficient uses the declared earlier predictor position and preserves the expected response residual."},{"criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"covered","evidence":["src/anomalyzer/relationships.py:152","tests/test_agentic.py:254","examples/agentic/lagged-response.json"],"notes":"Available and pending lineage retains existing source indexes and timestamp cutoffs; missing predictors do not become zero."},{"criterion":"When requested, the coefficient is fitted once from a declared training prefix and remains frozen.","status":"covered","evidence":["src/anomalyzer/relationships.py:47","tests/test_agentic.py:294","tests/test_agentic.py:330"],"notes":"The coefficient is fitted once after the complete declared prefix; the independent 20/7 result and prefix-invariance assertions pass."},{"criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"covered","evidence":["src/anomalyzer/relationships.py:122","tests/test_agentic.py:330"],"notes":"Short prefixes retain null derived values and explicit unavailable reasons, while an unidentifiable completed fit has a visible error."}],"ignore_findings":[],"done":true,"input_fingerprint":"fc4516ef6ab903dfa4918268d97049aff04fbf9556620ccb8259ff830ba85785","mode":"quick"},"reviewFindings":["Saved review is stale; reassess changed capability or implementation evidence."],"reviewStale":true,"health":"review","healthLabel":"Needs review","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","name":"Related-metric analysis","label":"Related-metric analysis / Investigate relationships I declare / Check delayed responses"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878},{"id":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","label":"Inspect related metrics be...","area":"relationships","scopes":["relationships","relationships/resolution"],"scope":"relationships/resolution","scopeLabel":"relationships/resolution","status":"implemented","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","acceptance":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"guidance":["Daily ratios and differences are supported. Daily lagged, normalized-residual and joint-condition definitions remain explicitly inapplicable; their native evidence remains eligible. Local-calendar aggregation is separate work.","Initial support should use explicit relationships and compatible boundaries; cross-cadence joining or automatic resampling needs a separate contract.","Proposed automated checks cover ratio-of-totals arithmetic, missing sources, mismatched period boundaries, and open-period evidence separation.","Manual review should inspect a conversion decline within an unfinished day and verify the explanation separates native evidence from partial totals."],"path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","dependencies":["data/quality/analyze-sparse-observations","relationships/comparisons/compare-declared-metrics"],"suggestedDependencies":[],"dependents":[],"implementationReferences":["src/anomalyzer/sparse_views.py","src/anomalyzer/relationships.py","src/anomalyzer/agent.py","tests/test_sparse.py","examples/sparse/unfinished_daily.json","docs/sparse-data.md"],"automatedChecks":[{"id":"ratio-of-sums","description":"Compute daily ratios from compatible sums and preserve missing-child coverage.","command":"python -m pytest -q tests/test_sparse.py::test_complete_daily_ratio_divides_sums_and_missing_child_invalidates_only_its_view"},{"id":"view-policies","description":"Keep native/daily rules separate and preserve case identity on replay.","command":"python -m pytest -q tests/test_sparse.py::test_daily_rules_are_not_silently_reused_from_hourly_rules tests/test_sparse.py::test_multi_resolution_mcp_drilldown_selects_correct_view_and_replays_case_ids"},{"id":"view-limits","description":"Report unsupported daily lag semantics without discarding native evidence.","command":"python -m pytest -q tests/test_sparse.py::test_unsupported_daily_relationship_is_explicit_and_does_not_reinterpret_lag"}],"manualChecks":["Inspect progression chapter 22 and confirm incomplete progress is not presented as a completed total or independent corroboration."],"verificationGaps":[],"review":{"depth":"partial","source":"deterministic-assessment","gaps":["Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."],"evidence":[],"intent_summary":"Implementation evidence synchronized from deterministic assessment for relationships/resolution/compare-finalized-subperiods.","criteria":[{"criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:22","src/anomalyzer/sparse_views.py:139","src/anomalyzer/sparse_views.py:143","src/anomalyzer/sparse_views.py:164","src/anomalyzer/agent.py:14","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:23","src/anomalyzer/agent.py:29","tests/test_sparse.py:18","tests/test_sparse.py:59","tests/test_sparse.py:87","tests/test_sparse.py:99","tests/test_sparse.py:117","docs/sparse-data.md:31","docs/sparse-data.md:127","docs/sparse-data.md:159","docs/sparse-data.md:199","docs/sparse-data.md:203"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:19","src/anomalyzer/sparse_views.py:75","src/anomalyzer/sparse_views.py:94","src/anomalyzer/sparse_views.py:122","src/anomalyzer/sparse_views.py:123","src/anomalyzer/relationships.py:7","src/anomalyzer/relationships.py:19","src/anomalyzer/relationships.py:33","src/anomalyzer/relationships.py:34","src/anomalyzer/relationships.py:36","src/anomalyzer/agent.py:34","tests/test_sparse.py:10","tests/test_sparse.py:40","tests/test_sparse.py:50","tests/test_sparse.py:60","tests/test_sparse.py:62","docs/sparse-data.md:17","docs/sparse-data.md:41","docs/sparse-data.md:68","docs/sparse-data.md:97","docs/sparse-data.md:99"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:70","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:170","src/anomalyzer/relationships.py:90","src/anomalyzer/relationships.py:173","tests/test_sparse.py:230","tests/test_sparse.py:259","tests/test_sparse.py:313","docs/sparse-data.md:41","docs/sparse-data.md:111","docs/sparse-data.md:194","docs/sparse-data.md:197","docs/sparse-data.md:198"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"criterion":"Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration.","status":"partial","evidence":["src/anomalyzer/sparse_views.py:1","src/anomalyzer/sparse_views.py:29","src/anomalyzer/sparse_views.py:82","src/anomalyzer/sparse_views.py:130","src/anomalyzer/sparse_views.py:159","src/anomalyzer/relationships.py:157","src/anomalyzer/relationships.py:192","src/anomalyzer/agent.py:20","src/anomalyzer/agent.py:47","tests/test_sparse.py:217","tests/test_sparse.py:247","tests/test_sparse.py:251","tests/test_sparse.py:264","tests/test_sparse.py:386","examples/sparse/unfinished_daily.json:208","docs/sparse-data.md:8","docs/sparse-data.md:11","docs/sparse-data.md:37","docs/sparse-data.md:41","docs/sparse-data.md:45"],"notes":"Related implementation text was found, but deterministic matching cannot prove the behavior. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"ignore_findings":[],"done":false},"reviewFindings":["review is not done","review depth is partial","Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration. Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit.","partial: Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","partial: Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","partial: Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","partial: Open-period progress remains separate from completed-period evidence and identifies coverage; correlated resolutions do not count as independent corroboration."],"reviewStale":false,"health":"gap","healthLabel":"Review findings","coverage":"partial","storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","name":"Related-metric analysis","label":"Related-metric analysis / Monitor the periods my business uses / Compare related metrics while the period is open"},"impact":0,"gaps":0,"r":38,"x":1270,"y":878}],"links":[{"source":"data/inputs/accept-analysis-requests","target":"analysis/evidence/calibrate-departures","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/cli/analyze-files-and-streams","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/identify-residual-patterns","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"analysis/expectations/model-seasonal-level-and-growth","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"interfaces/python/analyze-through-public-api","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"investigation/cases/group-related-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/expectations/reset-declared-regimes","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/analyze-sparse-observations","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/mcp/investigate-findings-and-evidence","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"investigation/assessments/describe-early-changes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"relationships/quality/require-compatible-sources","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/reset-declared-regimes","kind":"committed"},{"source":"data/quality/prepare-timestamped-series","target":"data/quality/exclude-incomplete-periods","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/mcp/connect-read-only-local-server","kind":"committed"},{"source":"investigation/assessments/describe-early-changes","target":"investigation/rules/check-declared-numerical-rules","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"relationships/comparisons/compare-declared-metrics","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/detection/report-point-anomaly-episodes","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/evidence/assess-reference-readiness","kind":"committed"},{"source":"analysis/evidence/calibrate-departures","target":"analysis/expectations/adapt-to-changing-growth","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/expectations/protect-reference-from-extremes","kind":"committed"},{"source":"data/quality/exclude-incomplete-periods","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"analysis/expectations/model-seasonal-level-and-growth","target":"analysis/resolution/monitor-within-unfinished-days","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"analysis/detection/identify-residual-patterns","target":"data/quality/identify-distinct-events","kind":"committed"},{"source":"data/inputs/accept-analysis-requests","target":"execution/resilience/retain-evidence-under-failure-and-limits","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"interfaces/cli/explain-findings-in-readable-output","kind":"committed"},{"source":"interfaces/python/analyze-through-public-api","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/mcp/connect-read-only-local-server","target":"interfaces/python/install-versioned-library","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/mcp/investigate-findings-and-evidence","target":"interfaces/skill/guide-anomaly-investigation","kind":"committed"},{"source":"interfaces/cli/analyze-files-and-streams","target":"interfaces/skill/visualize-analysis-on-request","kind":"committed"},{"source":"investigation/rules/check-declared-numerical-rules","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"investigation/cases/group-related-evidence","target":"investigation/policy/replay-action-eligibility","kind":"committed"},{"source":"relationships/quality/require-compatible-sources","target":"relationships/comparisons/evaluate-delayed-response","kind":"committed"},{"source":"data/quality/analyze-sparse-observations","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"},{"source":"relationships/comparisons/compare-declared-metrics","target":"relationships/resolution/compare-finalized-subperiods","kind":"committed"}],"scopes":["analysis","analysis/detection","analysis/evidence","analysis/expectations","analysis/rates","analysis/resolution","data","data/calendar","data/cohorts","data/inputs","data/quality","data/revisions","evaluation","evaluation/comparisons","evaluation/outcomes","evaluation/replay","execution","execution/incremental","execution/resilience","execution/state","interfaces","interfaces/cli","interfaces/export","interfaces/mcp","interfaces/python","interfaces/skill","investigation","investigation/assessments","investigation/cases","investigation/cohorts","investigation/history","investigation/policy","investigation/regimes","investigation/rules","relationships","relationships/cohorts","relationships/comparisons","relationships/quality","relationships/resolution"],"storyMaps":["Anomaly detection Skill","Basic CLI anomaly detection","Evidence and review policy","Future - Business-aware expectations","Future - Calendar and data readiness","Future - Cohort investigation","Future - Continuous analysis","Future - Historical evaluation","Future - Investigation memory","Incomplete-period monitoring","Python integration","Read-only MCP server","Related-metric analysis","Trend and seasonal analysis"],"hasUnassignedStoryMap":true};
 const width = 1440;
 const height = 980;
 const padding = 74;
diff --git a/.capabilities/interfaces/python/install-versioned-library.capability.yaml b/.capabilities/interfaces/python/install-versioned-library.capability.yaml
index 41933ad..6b2626b 100644
--- a/.capabilities/interfaces/python/install-versioned-library.capability.yaml
+++ b/.capabilities/interfaces/python/install-versioned-library.capability.yaml
@@ -72,9 +72,6 @@ agent:
       - Follow the release guide and confirm Guard retains domain severity and
         incident ownership; inspect readiness and evidence limits before
         adopting.
-    gaps:
-      - First immutable GitHub Release requires owner setup and a reviewed
-        version tag.
   review:
     depth: partial
     source: deterministic-assessment
@@ -112,26 +109,29 @@ agent:
           - scripts/release.py:39
           - scripts/release.py:48
           - scripts/release.py:54
-          - scripts/publish_release.py:19
-          - scripts/publish_release.py:27
-          - scripts/publish_release.py:29
-          - scripts/publish_release.py:39
-          - scripts/verify_lambda.sh:14
-          - scripts/verify_lambda.sh:17
-          - scripts/verify_lambda.sh:19
-          - scripts/verify_lambda.sh:25
-          - scripts/verify_lambda.sh:27
+          - scripts/publish_release.py:21
+          - scripts/publish_release.py:34
+          - scripts/publish_release.py:42
+          - scripts/publish_release.py:53
+          - scripts/verify_lambda.sh:15
+          - scripts/verify_lambda.sh:18
+          - scripts/verify_lambda.sh:20
+          - scripts/verify_lambda.sh:26
+          - scripts/verify_lambda.sh:28
+          - examples/library/guard_smoke.py:74
           - .github/workflows/ci.yml:14
           - .github/workflows/ci.yml:19
           - .github/workflows/ci.yml:27
-          - .github/workflows/ci.yml:41
-          - .github/workflows/ci.yml:46
-          - .github/workflows/release.yml:28
-          - docs/releases/README.md:12
-          - docs/releases/README.md:13
-          - docs/releases/README.md:14
+          - .github/workflows/ci.yml:43
+          - .github/workflows/ci.yml:48
+          - .github/workflows/release.yml:31
+          - .github/workflows/release.yml:52
+          - .github/workflows/release.yml:61
+          - docs/releases/README.md:16
           - docs/releases/README.md:17
           - docs/releases/README.md:18
+          - docs/releases/README.md:21
+          - docs/releases/README.md:22
           - tests/test_release.py:60
           - tests/test_release.py:71
           - tests/test_release.py:81
@@ -145,17 +145,19 @@ agent:
         status: partial
         evidence:
           - scripts/verify_distribution.py:1
-          - scripts/verify_distribution.py:42
+          - scripts/verify_distribution.py:48
           - examples/library/guard_smoke.py:32
           - examples/library/guard_smoke.py:49
           - examples/library/guard_smoke.py:65
           - examples/library/guard_smoke.py:69
-          - examples/library/guard_smoke.py:75
-          - .github/workflows/ci.yml:57
-          - docs/releases/README.md:13
-          - docs/releases/README.md:208
-          - docs/releases/README.md:211
-          - docs/releases/README.md:213
+          - examples/library/guard_smoke.py:76
+          - .github/workflows/ci.yml:59
+          - .github/workflows/release.yml:60
+          - docs/releases/README.md:17
+          - docs/releases/README.md:99
+          - docs/releases/README.md:217
+          - docs/releases/README.md:220
+          - docs/releases/README.md:222
         notes: Related implementation text was found, but deterministic matching cannot
           prove the behavior. Review the cited evidence. If it is correct, save
           agent.review evidence or make the criterion/reference wording more
@@ -173,27 +175,28 @@ agent:
           - scripts/release.py:85
           - scripts/release.py:92
           - scripts/publish_release.py:1
-          - scripts/publish_release.py:25
-          - scripts/publish_release.py:26
-          - scripts/publish_release.py:27
-          - scripts/publish_release.py:28
-          - scripts/verify_distribution.py:41
-          - scripts/verify_lambda.sh:24
+          - scripts/publish_release.py:18
+          - scripts/publish_release.py:19
+          - scripts/publish_release.py:20
+          - scripts/publish_release.py:40
+          - scripts/verify_distribution.py:47
+          - scripts/verify_lambda.sh:25
           - examples/library/guard_smoke.py:39
-          - .github/workflows/ci.yml:42
-          - .github/workflows/ci.yml:45
-          - .github/workflows/ci.yml:46
-          - .github/workflows/ci.yml:49
+          - .github/workflows/ci.yml:44
+          - .github/workflows/ci.yml:47
+          - .github/workflows/ci.yml:48
           - .github/workflows/ci.yml:51
-          - .github/workflows/release.yml:31
-          - .github/workflows/release.yml:33
+          - .github/workflows/ci.yml:53
+          - .github/workflows/release.yml:34
           - .github/workflows/release.yml:36
-          - .github/workflows/release.yml:37
-          - docs/releases/README.md:3
-          - docs/releases/README.md:12
-          - docs/releases/README.md:14
-          - docs/releases/README.md:15
+          - .github/workflows/release.yml:39
+          - .github/workflows/release.yml:40
+          - .github/workflows/release.yml:48
+          - docs/releases/README.md:8
+          - docs/releases/README.md:16
+          - docs/releases/README.md:18
           - docs/releases/README.md:19
+          - docs/releases/README.md:23
           - tests/test_release.py:1
           - tests/test_release.py:21
           - tests/test_release.py:23
@@ -211,21 +214,22 @@ agent:
           - pyproject.toml:33
           - scripts/release.py:81
           - scripts/release.py:82
-          - scripts/publish_release.py:40
-          - scripts/verify_distribution.py:41
+          - scripts/publish_release.py:54
+          - scripts/verify_distribution.py:47
           - scripts/verify_lambda.sh:6
-          - scripts/verify_lambda.sh:17
-          - scripts/verify_lambda.sh:19
-          - .github/workflows/ci.yml:30
-          - .github/workflows/ci.yml:31
-          - .github/workflows/ci.yml:51
-          - .github/workflows/ci.yml:52
-          - .github/workflows/ci.yml:90
-          - docs/releases/README.md:12
-          - docs/releases/README.md:15
-          - docs/releases/README.md:58
-          - docs/releases/README.md:59
-          - docs/releases/README.md:75
+          - scripts/verify_lambda.sh:18
+          - scripts/verify_lambda.sh:20
+          - .github/workflows/ci.yml:29
+          - .github/workflows/ci.yml:32
+          - .github/workflows/ci.yml:33
+          - .github/workflows/ci.yml:53
+          - .github/workflows/ci.yml:54
+          - .github/workflows/release.yml:62
+          - docs/releases/README.md:16
+          - docs/releases/README.md:19
+          - docs/releases/README.md:62
+          - docs/releases/README.md:63
+          - docs/releases/README.md:79
         notes: Related implementation text was found, but deterministic matching cannot
           prove the behavior. Review the cited evidence. If it is correct, save
           agent.review evidence or make the criterion/reference wording more
diff --git a/.capabilities/story-map-viewer.html b/.capabilities/story-map-viewer.html
index d54c2ec..7a0c90c 100644
--- a/.capabilities/story-map-viewer.html
+++ b/.capabilities/story-map-viewer.html
@@ -492,7 +492,7 @@ <h1>Anomalyzer story map</h1>
       <dialog id="detail-dialog"></dialog>
     </div>
     <script>
-const report = {"project":"Anomalyzer","capabilities":[{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","health":"action","coverage":"partial","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","references":{"total":11,"readable":11,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[{"code":"declared-gap","message":"interfaces/python/install-versioned-library: First immutable GitHub Release requires owner setup and a reviewed version tag.","capabilityId":"interfaces/python/install-versioned-library","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\python\\install-versioned-library.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Resolve or explicitly ignore verification gaps.","topFindings":[{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/publish_release.py","line":43,"excerpt":"\"Linux/Python 3.12 Lambda consumer verified offline. \""},{"reference":"scripts/verify_distribution.py","line":41,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":6,"excerpt":"bundle=\"$root/.tmp/lambda-consumer\""},{"reference":"scripts/verify_lambda.sh","line":18,"excerpt":"python -m pip download --only-binary=:all: --require-hashes \\"},{"reference":"scripts/verify_lambda.sh","line":20,"excerpt":"python -m pip install --no-index --only-binary=:all: --require-hashes \\"},{"reference":".github/workflows/ci.yml","line":29,"excerpt":"- run: uv run python examples/generate.py"},{"reference":".github/workflows/ci.yml","line":32,"excerpt":"- run: uv run python examples/run_demo.py"},{"reference":".github/workflows/ci.yml","line":33,"excerpt":"- run: uv run python examples/run_agentic_demo.py"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/ci.yml","line":54,"excerpt":"run: uv export --locked --no-dev --no-emit-project --output-file dist/runtime-dependencies.txt"},{"reference":"docs/releases/README.md","line":12,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":15,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":58,"excerpt":"uv run python examples/run_demo.py"},{"reference":"docs/releases/README.md","line":59,"excerpt":"uv run python examples/run_agentic_demo.py"},{"reference":"docs/releases/README.md","line":75,"excerpt":"versions, and tests a bundled Linux x86_64/Python 3.12 consumer in AWS's Lambda"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","evidence":[{"reference":"scripts/verify_distribution.py","line":1,"excerpt":"\"\"\"Install wheel AND sdist in clean environments outside the source tree (requires uv).\"\"\""},{"reference":"scripts/verify_distribution.py","line":42,"excerpt":"run([\"uv\", \"pip\", \"install\", \"--python\", python, \"--no-deps\", artifact], outside, env=env)"},{"reference":"examples/library/guard_smoke.py","line":32,"excerpt":"analyze_relationships, get_case, replay_policy,"},{"reference":"examples/library/guard_smoke.py","line":49,"excerpt":"assert legacy.methods[0].diagnostics[\"detection_readiness\"]"},{"reference":"examples/library/guard_smoke.py","line":65,"excerpt":"\"relationships\": [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"cost\","},{"reference":"examples/library/guard_smoke.py","line":69,"excerpt":"result = analyze_relationships(request)"},{"reference":"examples/library/guard_smoke.py","line":76,"excerpt":"short = analyze_relationships({**request.model_dump(mode=\"json\"),"},{"reference":".github/workflows/ci.yml","line":59,"excerpt":"- name: Test built wheel and sdist outside checkout"},{"reference":"docs/releases/README.md","line":13,"excerpt":"| Source distribution (`.tar.gz`, or sdist) | Source needed to build the package. Useful for rebuilding and inspection; installation also needs the declared build backend. |"},{"reference":"docs/releases/README.md","line":208,"excerpt":"`analyze_series`, `analyze_relationships`, `get_case`, `replay_policy` (also"},{"reference":"docs/releases/README.md","line":211,"excerpt":"`anomalyzer.contracts`. See the self-contained"},{"reference":"docs/releases/README.md","line":213,"excerpt":"all APIs, recipes, point/pattern evidence, relationships, cases and policy"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":34,"excerpt":"return hashlib.sha256(path.read_bytes()).hexdigest()"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/release.py","line":85,"excerpt":"suffix = f\" --hash=sha256:{digest(wheel)}\\n\""},{"reference":"scripts/release.py","line":92,"excerpt":"assets = sorted(p for p in dist.iterdir() if p.is_file()"},{"reference":"scripts/publish_release.py","line":1,"excerpt":"\"\"\"Publish tested assets without replacing tags or existing assets (GitHub CLI auth).\"\"\""},{"reference":"scripts/publish_release.py","line":25,"excerpt":"remote_sha = json.loads(gh(\"api\", f\"repos/{REPOSITORY}/commits/{tag}\"))[\"sha\"]"},{"reference":"scripts/publish_release.py","line":26,"excerpt":"if remote_sha != revision:"},{"reference":"scripts/publish_release.py","line":27,"excerpt":"raise ValueError(\"Remote tag no longer points to the tested source revision\")"},{"reference":"scripts/publish_release.py","line":28,"excerpt":"endpoint = f\"repos/{REPOSITORY}/releases/tags/{tag}\""},{"reference":"scripts/verify_distribution.py","line":41,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":25,"excerpt":"sha256sum *.whl > /output/wheelhouse-SHA256SUMS"},{"reference":"examples/library/guard_smoke.py","line":39,"excerpt":"assert anomalyzer.__version__ == importlib.metadata.version(\"anomalyzer\")"},{"reference":".github/workflows/ci.yml","line":44,"excerpt":"- name: Check release tag before building"},{"reference":".github/workflows/ci.yml","line":47,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/ci.yml","line":48,"excerpt":"run: python scripts/release.py check-version --tag \"$RELEASE_TAG\""},{"reference":".github/workflows/ci.yml","line":51,"excerpt":"- name: Check package metadata"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/release.yml","line":31,"excerpt":"name: release-dist-${{ github.sha }}"},{"reference":".github/workflows/release.yml","line":33,"excerpt":"- name: Stage tested assets for owner publication"},{"reference":".github/workflows/release.yml","line":36,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/release.yml","line":37,"excerpt":"run: python scripts/publish_release.py --tag \"$RELEASE_TAG\" --revision \"$GITHUB_SHA\""},{"reference":"docs/releases/README.md","line":3,"excerpt":"Anomalyzer is a normal Python dependency. Version **0.2.0** is the first release"},{"reference":"docs/releases/README.md","line":12,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":14,"excerpt":"| Version tag (`v0.2.0`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |"},{"reference":"docs/releases/README.md","line":15,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"it for package metadata and the public `__version__` imports it."},{"reference":"tests/test_release.py","line":1,"excerpt":"\"\"\"Release safety: reject wrong tags and refuse changes to published artifacts.\"\"\""},{"reference":"tests/test_release.py","line":21,"excerpt":"def test_release_refuses_mismatched_or_unsupported_tag(tag):"},{"reference":"tests/test_release.py","line":23,"excerpt":"release.check_tag(tag)"},{"reference":"tests/test_release.py","line":52,"excerpt":"def test_published_matching_release_is_noop(monkeypatch, tmp_path):"},{"reference":"tests/test_release.py","line":81,"excerpt":"\"version\": release.version(), \"source_revision\": \"a\" * 40, \"tag\": f\"v{release.version()}\","}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","order":20}},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","health":"review","coverage":"partial","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Moving-range findings identify variation changes separately from location changes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","order":60}},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","health":"review","coverage":"partial","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","order":40}},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","health":"review","coverage":"partial","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","order":20}},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","health":"review","coverage":"partial","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","order":30}},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","health":"review","coverage":"partial","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","order":40}},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","health":"review","coverage":"partial","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","order":10}},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","health":"review","coverage":"partial","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","order":30}},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","health":"review","coverage":"partial","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","order":50}},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/select-season-length.capability.yaml","health":"review","coverage":"partial","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"An explicit season length is honored and reported as explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","order":20}},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","status":"implemented","area":"analysis","path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","health":"review","coverage":"partial","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","order":20}},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","status":"implemented","area":"data","path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","health":"review","coverage":"partial","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","order":10}},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","status":"implemented","area":"data","path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","health":"review","coverage":"partial","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","references":{"total":10,"readable":10,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":13,"excerpt":"\"seasonal-residual-v1\", \"adaptive-seasonal-v1\", \"multi-resolution-v1\","},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":297,"excerpt":"if (self.data_policy or self.coverage or self.available_at or self.revisions) and self.timestamps is None:"},{"reference":"src/anomalyzer/contracts.py","line":298,"excerpt":"raise ValueError(\"source policies/coverage/revisions require original timestamps\")"},{"reference":"src/anomalyzer/contracts.py","line":437,"excerpt":"total_points = sum(len(dataset.values) + len(dataset.revisions) for dataset in self.datasets)"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":64,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse.py","line":66,"excerpt":"if coverage != \"complete\" or previous[\"coverage\"] != \"complete\" or ds.available_at or ds.revisions:"},{"reference":"src/anomalyzer/sparse.py","line":83,"excerpt":"observed_at = timestamp(revision.timestamp, ds.timezone)"},{"reference":"src/anomalyzer/sparse_views.py","line":1,"excerpt":"\"\"\"Schema-1.1 native and complete-day evidence with independent view eligibility.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":28,"excerpt":"case.revision_id = f\"{resolution}:{case.revision_id}\""},{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/orchestrator.py","line":106,"excerpt":"if request.config.recipe == \"multi-resolution-v1\":"},{"reference":"src/anomalyzer/orchestrator.py","line":107,"excerpt":"from .sparse_views import analyze_multi_request"},{"reference":"src/anomalyzer/orchestrator.py","line":133,"excerpt":"if \"data_policy\" in source.quality and request.config.recipe != \"multi-resolution-v1\""},{"reference":"src/anomalyzer/orchestrator.py","line":166,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":180,"excerpt":"reset_positions=active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of),"},{"reference":"src/anomalyzer/relationships.py","line":160,"excerpt":"source_timestamps = {source.dataset.id: source.times[source_index]"},{"reference":"src/anomalyzer/relationships.py","line":162,"excerpt":"cutoff = max(source_timestamps.values())"},{"reference":"src/anomalyzer/relationships.py","line":166,"excerpt":"source_indexes=source_indexes, source_timestamps=source_timestamps,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":124,"excerpt":"def test_sparse_phase_without_support_is_explicitly_unavailable():"},{"reference":"tests/test_sparse.py","line":181,"excerpt":"ds[\"timestamps\"][1] = \"2026-02-01T00:00:00Z\""},{"reference":"docs/sparse-data.md","line":30,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":87,"excerpt":"otherwise bounds come from the supplied as-of-available timestamp range."},{"reference":"docs/sparse-data.md","line":105,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"examples/sparse_examples.py","line":59,"excerpt":"daily = request(dataset([1.0]*60, data_policy={\"measurement\":\"additive_volume\"}), recipe=\"multi-resolution-v1\","},{"reference":"examples/sparse_examples.py","line":95,"excerpt":"daily = next(item for item in outputs[\"unfinished_daily\"].dataset_results if item.resolution == \"completed_period\")"},{"reference":"examples/render_readme_progression.py","line":686,"excerpt":"def figure_multi_resolution():"},{"reference":"examples/render_readme_progression.py","line":699,"excerpt":"\"config\": {\"recipe\": \"multi-resolution-v1\", \"aggregate_function\": \"sum\"},"},{"reference":"examples/render_readme_progression.py","line":710,"excerpt":"assert partial[\"included_in_completed_period_view\"] is False"},{"reference":"examples/render_readme_progression.py","line":767,"excerpt":"complete_count = result.data_quality[\"views\"][\"completed_period\"][\"observation_count\"]"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":51,"excerpt":"def valid_reset_points(cls, values):"},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":71,"excerpt":"def numeric_values(cls, data):"},{"reference":"src/anomalyzer/contracts.py","line":85,"excerpt":"if self.period_statuses is not None and len(self.period_statuses) != len(self.values):"},{"reference":"src/anomalyzer/contracts.py","line":86,"excerpt":"raise ValueError(\"period_statuses and values must have equal length\")"},{"reference":"src/anomalyzer/sparse.py","line":3,"excerpt":"Calendar coordinates remain dense only inside an explicitly bounded input grid."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":42,"excerpt":"\"\"\"Return the explicit grid, eligible model values and inspectable provenance.\"\"\""},{"reference":"src/anomalyzer/sparse.py","line":48,"excerpt":"statuses = ds.period_statuses or [\"complete\"] * len(ds.values)"},{"reference":"src/anomalyzer/sparse.py","line":54,"excerpt":"\"complete\" if value is not None else \"empty\" if policy.absence_means_zero else \"unknown\")"},{"reference":"src/anomalyzer/sparse_views.py","line":72,"excerpt":"measurements = quality[\"rule_values\"]"},{"reference":"src/anomalyzer/sparse_views.py","line":75,"excerpt":"if policy.observation_basis != \"calendar\":"},{"reference":"src/anomalyzer/sparse_views.py","line":82,"excerpt":"if policy.measurement != \"additive_volume\" or cfg.aggregate_function != \"sum\":"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":98,"excerpt":"daily_times, daily_values, coverage, records, available_at = [], [], [], [], []"},{"reference":"src/anomalyzer/orchestrator.py","line":28,"excerpt":"if not any(row.get(\"available_at\") and (row.get(\"value\") is not None or row.get(\"coverage\") == \"excluded\") for row in samples):"},{"reference":"src/anomalyzer/orchestrator.py","line":63,"excerpt":"unavailable values. Dropping only that leading run lets the ordinary engine"},{"reference":"src/anomalyzer/orchestrator.py","line":127,"excerpt":"prepared_points += len(source.values) + len(dataset.revisions)"},{"reference":"src/anomalyzer/orchestrator.py","line":139,"excerpt":"source.quality.get(\"rule_values\", source.values)[:source.quality.get(\"grid_count\", source.quality[\"observation_count\"])], dataset.rules,"},{"reference":"src/anomalyzer/orchestrator.py","line":191,"excerpt":"[dict(value=p.value, available_at=p.latest_source_cutoff if any(p.source_available_at.values()) else None) for p in lineage])"},{"reference":"src/anomalyzer/relationships.py","line":36,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/relationships.py","line":58,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":98,"excerpt":"value, unavailable = None, \"zero_denominator\""},{"reference":"src/anomalyzer/relationships.py","line":131,"excerpt":"y, x = response.values[index], predictor.values[predictor_index]"},{"reference":"src/anomalyzer/relationships.py","line":141,"excerpt":"source_values = [source.values[source_index]"},{"reference":"src/anomalyzer/early.py","line":58,"excerpt":"for index, (observed_at, value) in enumerate(zip(times, values)):"},{"reference":"src/anomalyzer/early.py","line":137,"excerpt":"criterion_met and policy.allow_explicit_rules)"},{"reference":"tests/test_sparse.py","line":45,"excerpt":"def test_zero_is_only_authorized_for_complete_empty_additive_volume():"},{"reference":"tests/test_sparse.py","line":46,"excerpt":"request = body([2, None, None, 0, 9], data_policy={\"missing\": \"zero_for_empty\", \"measurement\": \"additive_volume\"},"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":52,"excerpt":"assert rows[1][\"reason\"] == \"confirmed_empty_additive_volume\""},{"reference":"tests/test_sparse.py","line":58,"excerpt":"def test_global_collection_contract_authorizes_omitted_slots_but_not_explicit_unknown():"},{"reference":"docs/sparse-data.md","line":4,"excerpt":"Schema 1.1 therefore preserves missing values by default. An expectation can be"},{"reference":"docs/sparse-data.md","line":33,"excerpt":"data_policy=DataPolicy(),  # preserve missing; hour 2 is also unknown"},{"reference":"docs/sparse-data.md","line":41,"excerpt":"| Numeric value, including zero, `coverage=\"complete\"` | A measurement eligible for rules and model support. Explicit zero remains zero. |"},{"reference":"docs/sparse-data.md","line":43,"excerpt":"| `coverage=\"empty\"` | Caller confirms a finalized period with no reported volume. Preserve as missing unless an additive zero policy explicitly authorizes conversion. |"},{"reference":"docs/sparse-data.md","line":44,"excerpt":"| `coverage=\"unknown\"` | Collection/exposure is unknown; never filled even under a global empty-volume contract. |"},{"reference":"examples/sparse_examples.py","line":22,"excerpt":"preserve = request(dataset(values, coverage=[\"empty\" if v is None else \"complete\" for v in values],"},{"reference":"examples/sparse_examples.py","line":23,"excerpt":"data_policy={\"measurement\": \"additive_volume\"}))"},{"reference":"examples/sparse_examples.py","line":25,"excerpt":"zero[\"datasets\"][0][\"data_policy\"][\"missing\"] = \"zero_for_empty\""},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":28,"excerpt":"coverage=[\"complete\"]*2+[\"unknown\"]+[\"complete\"]*6+[\"unknown\"]+[\"complete\"]*3+[\"incomplete\"], data_policy={}))"},{"reference":"examples/render_readme_progression.py","line":135,"excerpt":"def chart(svg, values, overlays, x, y, width, height, *, y_bounds=None,"},{"reference":"examples/render_readme_progression.py","line":139,"excerpt":"all_values = [values] + [series for _, series, *_ in overlays]"},{"reference":"examples/render_readme_progression.py","line":140,"excerpt":"low, high = y_bounds or bounds(*all_values)"},{"reference":"examples/render_readme_progression.py","line":166,"excerpt":"if value is not None and (index == 0 or values[index-1] is None) and (index == count-1 or values[index+1] is None):"},{"reference":"examples/render_readme_progression.py","line":206,"excerpt":"values = [80 + 0.65 * index + [0, 0.25, -0.15, 0.10, -0.20][index % 5]"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":242,"excerpt":"missing: Literal[\"preserve\", \"zero_for_empty\"] = \"preserve\""},{"reference":"src/anomalyzer/contracts.py","line":245,"excerpt":"observation_basis: Literal[\"calendar\", \"event_index\"] = \"calendar\""},{"reference":"src/anomalyzer/contracts.py","line":258,"excerpt":"if self.observation_basis == \"event_index\" and ("},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":49,"excerpt":"for index, (text, value, status) in enumerate(zip(ds.timestamps, ds.values, statuses)):"},{"reference":"src/anomalyzer/sparse.py","line":57,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":63,"excerpt":"if cfg.duplicate_policy == \"error\" or policy.observation_basis == \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":64,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":128,"excerpt":"policy_zero_children=sum(quality[\"samples\"][i][\"reason\"] == \"confirmed_empty_additive_volume\" for i in indexes),"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":134,"excerpt":"timestamps=daily_times, values=daily_values, frequency=\"1d\", coverage=coverage, available_at=available_at,"},{"reference":"src/anomalyzer/orchestrator.py","line":37,"excerpt":"assessment.baseline[\"known_input_availability_cutoff\"] = cutoffs[assessment.index]"},{"reference":"src/anomalyzer/orchestrator.py","line":38,"excerpt":"assessment.assumptions.append(\"Availability cutoff is conservative across known prior inputs; unknown arrival history cannot establish delivery-time detection performance.\")"},{"reference":"src/anomalyzer/orchestrator.py","line":166,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":178,"excerpt":"relationship.id, derived.timestamps or [], derived.values,"},{"reference":"src/anomalyzer/orchestrator.py","line":180,"excerpt":"reset_positions=active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of),"},{"reference":"src/anomalyzer/relationships.py","line":40,"excerpt":"raise ValueError(\"relationship timestamps cannot be positional\")"},{"reference":"src/anomalyzer/relationships.py","line":57,"excerpt":"for index in range(relationship.lag, coefficient_cutoff + 1):"},{"reference":"src/anomalyzer/relationships.py","line":58,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":96,"excerpt":"if relationship.zero_denominator == \"error\":"},{"reference":"src/anomalyzer/relationships.py","line":97,"excerpt":"raise ValueError(f\"zero denominator at aligned index {index}\")"},{"reference":"src/anomalyzer/early.py","line":140,"excerpt":"unavailable_rules = [dict(id=rule.id, reason=\"no_comparable_prior\" if previous is None else \"relative_change_from_zero\")"},{"reference":"src/anomalyzer/early.py","line":147,"excerpt":"target_id=target_id, index=index, timestamp=observed_at,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":68,"excerpt":"{\"observation_basis\": \"event_index\", \"grid_start\": \"2026-01-01T00:00:00Z\"},"},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":129,"excerpt":"assert any(row[\"reason\"] == \"no_training_support_for_calendar_phase\" for row in result.methods[0].diagnostics[\"unavailable\"])"},{"reference":"docs/sparse-data.md","line":30,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":105,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"docs/sparse-data.md","line":110,"excerpt":"rejected. Event-index timestamps must be unique; ties are refused rather than"},{"reference":"docs/sparse-data.md","line":141,"excerpt":"original calendar indexes and timestamps. Post-gap learning requires an explicit"},{"reference":"docs/sparse-data.md","line":155,"excerpt":"## Comparable event-index observations"},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":38,"excerpt":"events[\"timestamps\"] = [(datetime(2026,1,1,tzinfo=timezone.utc)+timedelta(hours=h)).isoformat() for h in hours]"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":81,"excerpt":"preserve, zero = [outputs[n].dataset_results[0] for n in (\"preserve\", \"zero_empty\")]"},{"reference":"examples/render_readme_progression.py","line":654,"excerpt":"svg.text(430, 467, \"Sample index (zero-based)\", 15, MUTED, anchor=\"middle\")"},{"reference":"examples/render_readme_progression.py","line":697,"excerpt":"\"timestamps\": [(origin + timedelta(hours=index)).isoformat()"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"},{"reference":"examples/render_readme_progression.py","line":887,"excerpt":"shared_high = max(v for v in observed(\"preserve\")+expectation(\"preserve\")+observed(\"zero_empty\")+expectation(\"zero_empty\") if v is not None)*1.12"},{"reference":"examples/render_readme_progression.py","line":897,"excerpt":"header(svg, \"15\", \"Keep zero, unknown and unfinished distinct\", \"Blue zeros are actual measurements; gaps stay gaps; the orange open-period value is excluded.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","order":30}},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","status":"implemented","area":"data","path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","health":"review","coverage":"partial","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","order":10}},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","status":"implemented","area":"data","path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","health":"review","coverage":"partial","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","order":20}},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","status":"implemented","area":"execution","path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","health":"review","coverage":"partial","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","order":50}},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","health":"review","coverage":"partial","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","order":50}},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","health":"review","coverage":"partial","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Readiness status remains visible, including when it was not assessed.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","order":60}},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","health":"review","coverage":"partial","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Analysis tool calls return structured content together with compact text.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","order":10}},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","order":20}},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","health":"review","coverage":"partial","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","order":10}},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","health":"review","coverage":"partial","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","order":10}},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","health":"review","coverage":"partial","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","order":20}},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","status":"implemented","area":"investigation","path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","health":"review","coverage":"partial","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","order":10}},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","status":"implemented","area":"investigation","path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","order":40}},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","status":"implemented","area":"investigation","path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","health":"review","coverage":"partial","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Default policy leaves action and notification eligibility disabled.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","order":60}},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","status":"implemented","area":"investigation","path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","health":"review","coverage":"partial","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","order":30}},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","health":"review","coverage":"partial","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","order":20}},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","health":"review","coverage":"partial","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","order":30}},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","status":"implemented","area":"relationships","path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","health":"review","coverage":"partial","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Difference relationships require identical declared source units and matching output units.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","order":10}},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","status":"implemented","area":"relationships","path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","health":"review","coverage":"partial","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":22,"excerpt":"cases = compose_cases([d for d in datasets if d.resolution == resolution],"},{"reference":"src/anomalyzer/sparse_views.py","line":146,"excerpt":"native_sources = [d.model_copy(update={\"data_policy\": d.data_policy or DataPolicy()}) for d in request.datasets]"},{"reference":"src/anomalyzer/sparse_views.py","line":149,"excerpt":"native_request = request.model_copy(update={\"datasets\": native_sources, \"config\": native_config})"},{"reference":"src/anomalyzer/sparse_views.py","line":169,"excerpt":"daily_request = request.model_copy(update={\"datasets\": sources, \"config\": daily_config, \"relationships\": daily_relationships})"},{"reference":"src/anomalyzer/agent.py","line":14,"excerpt":"\"\"\"Use legacy semantics for a Dataset; explicit V11 fields select source policies.\"\"\""},{"reference":"src/anomalyzer/agent.py","line":20,"excerpt":"name in dataset for name in (\"data_policy\", \"coverage\", \"available_at\", \"revisions\", \"rules\", \"completed_period_rules\")):"},{"reference":"src/anomalyzer/agent.py","line":23,"excerpt":"return analyze({\"schema_version\": \"1.1\", \"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"src/anomalyzer/agent.py","line":29,"excerpt":"return analyze({\"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"tests/test_sparse.py","line":18,"excerpt":"return {\"schema_version\": \"1.1\", \"datasets\": [{\"id\": \"series\", \"entity\": {\"customer\": \"test\"},"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":88,"excerpt":"def test_declared_duplicate_aggregation_preserves_provenance(policy, value):"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":118,"excerpt":"changed[\"datasets\"][0][\"values\"][-10:] = [999]*10"},{"reference":"docs/sparse-data.md","line":35,"excerpt":"result = analyze({\"schema_version\": \"1.1\", \"datasets\": [source.model_dump(mode=\"json\")]})"},{"reference":"docs/sparse-data.md","line":136,"excerpt":"incomplete, excluded or reset boundaries, regardless of point-reference policy."},{"reference":"docs/sparse-data.md","line":142,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":172,"excerpt":"basis. A caller-declared maximum time separation can limit change comparisons."},{"reference":"docs/sparse-data.md","line":218,"excerpt":"Daily lagged, normalized-residual and joint-condition definitions remain"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":70,"excerpt":"\"\"\"Sum only wholly observed finalized additive children. Never average ratios.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":175,"excerpt":"applicability=\"completed-day view supports ratio-of-sums and differences of compatible additive sums only\","},{"reference":"src/anomalyzer/relationships.py","line":90,"excerpt":"used = [(numerator, index), (denominator, index)]"},{"reference":"src/anomalyzer/relationships.py","line":173,"excerpt":"\"complete\" if source.values[i] is not None else \"unknown_or_incomplete\" for source, i in used}"},{"reference":"tests/test_sparse.py","line":231,"excerpt":"request[\"relationships\"] = [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"series\", \"denominator\": \"calls\", \"units\": \"dollars_per_call\"}]"},{"reference":"tests/test_sparse.py","line":260,"excerpt":"def test_daily_rules_are_not_silently_reused_from_hourly_rules():"},{"reference":"tests/test_sparse.py","line":279,"excerpt":"def test_adaptive_gap_does_not_silently_train_a_post_gap_episode():"},{"reference":"tests/test_sparse.py","line":322,"excerpt":"def test_restart_policy_cannot_reuse_a_prior_regime_reference():"},{"reference":"tests/test_sparse.py","line":431,"excerpt":"assert row[\"complete\"] and row[\"aggregate_value\"] == 24 and not row[\"model_eligible\"]"},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":120,"excerpt":"It then collects `calibration_size` eligible forecast residuals, not elapsed slots."},{"reference":"docs/sparse-data.md","line":142,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":213,"excerpt":"partial sums are progress metadata, never a completed total. Empty days are not"},{"reference":"docs/sparse-data.md","line":216,"excerpt":"Daily ratios divide summed compatible numerator and denominator quantities;"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":130,"excerpt":"source_grid_indexes=indexes, observed_sum=(math.fsum(measurements[i] for i in indexes if measurements[i] is not None)"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":36,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/agent.py","line":34,"excerpt":"def analyze_relationships(request: RequestV11 | dict) -> ResultV11:"},{"reference":"tests/test_sparse.py","line":10,"excerpt":"from anomalyzer import analyze, analyze_series, analyze_relationships, get_case, replay_policy"},{"reference":"tests/test_sparse.py","line":41,"excerpt":"assert right.data_quality[\"samples\"][8][\"source_index\"] == -1"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":61,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, 0, 0, 2]"},{"reference":"tests/test_sparse.py","line":63,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, None, 0, 2]"},{"reference":"docs/sparse-data.md","line":21,"excerpt":"works through `analyze`, `analyze_relationships`, CLI JSON and read-only MCP."},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":72,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":76,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":106,"excerpt":"retains its unique-source requirement. With an explicit `data_policy`, complete,"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","order":30}},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","status":"planned","area":"analysis","path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","health":"planned","coverage":"uncovered","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","order":10}},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","status":"planned","area":"analysis","path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","health":"planned","coverage":"uncovered","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","order":20}},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","status":"planned","area":"data","path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","health":"planned","coverage":"uncovered","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","order":10}},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","status":"planned","area":"data","path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","health":"planned","coverage":"uncovered","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","order":10}},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","status":"planned","area":"data","path":".capabilities/data/quality/report-source-freshness.capability.yaml","health":"planned","coverage":"uncovered","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","order":20}},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","status":"planned","area":"data","path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","order":30}},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","status":"planned","area":"evaluation","path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","order":30}},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","status":"planned","area":"evaluation","path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","health":"planned","coverage":"uncovered","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","order":20}},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","status":"planned","area":"evaluation","path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","order":10}},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","status":"planned","area":"execution","path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","order":20}},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","status":"planned","area":"execution","path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","order":10}},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","status":"planned","area":"interfaces","path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","order":20}},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","status":"planned","area":"investigation","path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","order":20}},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","order":10}},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","health":"planned","coverage":"uncovered","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","order":30}},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","status":"planned","area":"investigation","path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","order":30}},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","status":"planned","area":"relationships","path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","order":30}}],"byStoryMap":{"unassigned":[],"releases":[{"release":"Anomaly detection Skill","capabilities":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","health":"review","coverage":"partial","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","order":10}},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","health":"review","coverage":"partial","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","order":20}}],"deliveryStrategy":{"release":"Anomaly detection Skill","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/skill/guide-anomaly-investigation"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/skill/guide-anomaly-investigation needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Ask an agent to investigate a change"],"missingBackbones":[],"stepCoverageGaps":["Ask an agent to investigate a change > Inspect a requested chart"],"rationale":"Starts with the earliest mapped step in Ask an agent to investigate a change."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/skill/visualize-analysis-on-request"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/skill/visualize-analysis-on-request needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Ask an agent to investigate a change"],"missingBackbones":[],"stepCoverageGaps":["Ask an agent to investigate a change > Get an evidence-grounded explanation"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/skill/visualize-analysis-on-request."}]},"presentation":{"outcome":"Anomaly detection Skill helps teams move from Get an evidence-grounded explanation to Inspect a requested chart across 1 backbone activity.","narrativePath":[{"backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","capabilityIds":["interfaces/skill/guide-anomaly-investigation"],"health":"review"},{"backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","capabilityIds":["interfaces/skill/visualize-analysis-on-request"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Ask an agent to investigate a change > Get an evidence-grounded explanation","message":"Ask an agent to investigate a change > Get an evidence-grounded explanation has needs-review coverage."},{"kind":"weak","label":"Ask an agent to investigate a change > Inspect a requested chart","message":"Ask an agent to investigate a change > Inspect a requested chart has needs-review coverage."}]}},{"release":"Basic CLI anomaly detection","capabilities":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","health":"review","coverage":"partial","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","order":40}},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","health":"review","coverage":"partial","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","order":30}},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","status":"implemented","area":"data","path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","health":"review","coverage":"partial","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","order":10}},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","status":"implemented","area":"data","path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","health":"review","coverage":"partial","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Duplicate timestamps are rejected unless an aggregation policy is supplied; duplicate counts and transformations remain visible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","order":20}},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","health":"review","coverage":"partial","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","order":50}},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","health":"review","coverage":"partial","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Readiness status remains visible, including when it was not assessed.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","order":60}}],"deliveryStrategy":{"release":"Basic CLI anomaly detection","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/inputs/accept-analysis-requests"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/inputs/accept-analysis-requests needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Analyze a local series"],"missingBackbones":[],"stepCoverageGaps":["Analyze a local series > Establish evidence support","Analyze a local series > Find spikes and drops","Analyze a local series > Prepare the comparison","Analyze a local series > Read the findings","Analyze a local series > Run from files or stdin"],"rationale":"Starts with the earliest mapped step in Analyze a local series."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["data/quality/prepare-timestamped-series","analysis/evidence/calibrate-departures","analysis/detection/report-point-anomaly-episodes","interfaces/cli/analyze-files-and-streams","interfaces/cli/explain-findings-in-readable-output"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/detection/report-point-anomaly-episodes needs review; analysis/evidence/calibrate-departures needs review; data/quality/prepare-timestamped-series needs review; interfaces/cli/analyze-files-and-streams needs review; interfaces/cli/explain-findings-in-readable-output needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Analyze a local series"],"missingBackbones":[],"stepCoverageGaps":["Analyze a local series > Supply a numeric series"],"rationale":"Prioritizes capabilities with delivery risk signals: data/quality/prepare-timestamped-series, analysis/evidence/calibrate-departures, analysis/detection/report-point-anomaly-episodes, interfaces/cli/analyze-files-and-streams, interfaces/cli/explain-findings-in-readable-output."}]},"presentation":{"outcome":"Basic CLI anomaly detection helps teams move from Supply a numeric series to Read the findings across 1 backbone activity.","narrativePath":[{"backbone":"Analyze a local series","step":"Supply a numeric series","capabilityIds":["data/inputs/accept-analysis-requests"],"health":"review"},{"backbone":"Analyze a local series","step":"Prepare the comparison","capabilityIds":["data/quality/prepare-timestamped-series"],"health":"review"},{"backbone":"Analyze a local series","step":"Establish evidence support","capabilityIds":["analysis/evidence/calibrate-departures"],"health":"review"},{"backbone":"Analyze a local series","step":"Find spikes and drops","capabilityIds":["analysis/detection/report-point-anomaly-episodes"],"health":"review"},{"backbone":"Analyze a local series","step":"Run from files or stdin","capabilityIds":["interfaces/cli/analyze-files-and-streams"],"health":"review"},{"backbone":"Analyze a local series","step":"Read the findings","capabilityIds":["interfaces/cli/explain-findings-in-readable-output"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Analyze a local series > Establish evidence support","message":"Analyze a local series > Establish evidence support has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Find spikes and drops","message":"Analyze a local series > Find spikes and drops has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Prepare the comparison","message":"Analyze a local series > Prepare the comparison has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Read the findings","message":"Analyze a local series > Read the findings has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Run from files or stdin","message":"Analyze a local series > Run from files or stdin has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Supply a numeric series","message":"Analyze a local series > Supply a numeric series has needs-review coverage."}]}},{"release":"Evidence and review policy","capabilities":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","health":"review","coverage":"partial","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","order":20}},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","status":"implemented","area":"execution","path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","health":"review","coverage":"partial","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","order":50}},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","status":"implemented","area":"investigation","path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","health":"review","coverage":"partial","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","order":10}},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","status":"implemented","area":"investigation","path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","order":40}},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","status":"implemented","area":"investigation","path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","health":"review","coverage":"partial","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Default policy leaves action and notification eligibility disabled.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","order":60}},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","status":"implemented","area":"investigation","path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","health":"review","coverage":"partial","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","order":30}}],"deliveryStrategy":{"release":"Evidence and review policy","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["investigation/assessments/describe-early-changes"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"investigation/assessments/describe-early-changes needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Decide how much to rely on a finding"],"missingBackbones":[],"stepCoverageGaps":["Decide how much to rely on a finding > Apply my stated limits","Decide how much to rely on a finding > Assess baseline quality","Decide how much to rely on a finding > Inspect incomplete execution","Decide how much to rely on a finding > Investigate related evidence","Decide how much to rely on a finding > Review action eligibility"],"rationale":"Starts with the earliest mapped step in Decide how much to rely on a finding."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/evidence/assess-reference-readiness","investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence","execution/resilience/retain-evidence-under-failure-and-limits","investigation/policy/replay-action-eligibility"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/evidence/assess-reference-readiness needs review; execution/resilience/retain-evidence-under-failure-and-limits needs review; investigation/cases/group-related-evidence needs review; investigation/policy/replay-action-eligibility needs review; investigation/rules/check-declared-numerical-rules needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Decide how much to rely on a finding"],"missingBackbones":[],"stepCoverageGaps":["Decide how much to rely on a finding > Inspect early evidence"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/evidence/assess-reference-readiness, investigation/rules/check-declared-numerical-rules, investigation/cases/group-related-evidence, execution/resilience/retain-evidence-under-failure-and-limits, investigation/policy/replay-action-eligibility."}]},"presentation":{"outcome":"Evidence and review policy helps teams move from Inspect early evidence to Review action eligibility across 1 backbone activity.","narrativePath":[{"backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","capabilityIds":["investigation/assessments/describe-early-changes"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","capabilityIds":["analysis/evidence/assess-reference-readiness"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","capabilityIds":["investigation/rules/check-declared-numerical-rules"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","capabilityIds":["investigation/cases/group-related-evidence"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","capabilityIds":["execution/resilience/retain-evidence-under-failure-and-limits"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Review action eligibility","capabilityIds":["investigation/policy/replay-action-eligibility"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Decide how much to rely on a finding > Apply my stated limits","message":"Decide how much to rely on a finding > Apply my stated limits has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Assess baseline quality","message":"Decide how much to rely on a finding > Assess baseline quality has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Inspect early evidence","message":"Decide how much to rely on a finding > Inspect early evidence has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Inspect incomplete execution","message":"Decide how much to rely on a finding > Inspect incomplete execution has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Investigate related evidence","message":"Decide how much to rely on a finding > Investigate related evidence has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Review action eligibility","message":"Decide how much to rely on a finding > Review action eligibility has needs-review coverage."}]}},{"release":"Future - Business-aware expectations","capabilities":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","status":"planned","area":"analysis","path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","health":"planned","coverage":"uncovered","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","order":10}},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","status":"planned","area":"analysis","path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","health":"planned","coverage":"uncovered","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","order":20}},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","status":"planned","area":"investigation","path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","order":30}}],"deliveryStrategy":{"release":"Future - Business-aware expectations","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["analysis/expectations/model-multiple-seasonal-cycles"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"analysis/expectations/model-multiple-seasonal-cycles has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Make expectations fit the process"],"missingBackbones":[],"stepCoverageGaps":["Make expectations fit the process > Judge rates using their exposure","Make expectations fit the process > Review a proposed regime transition"],"rationale":"Starts with the earliest mapped step in Make expectations fit the process."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["analysis/rates/assess-bounded-count-rates","investigation/regimes/propose-reviewed-baseline-changes"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"analysis/rates/assess-bounded-count-rates has verification gaps; investigation/regimes/propose-reviewed-baseline-changes has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Make expectations fit the process"],"missingBackbones":[],"stepCoverageGaps":["Make expectations fit the process > Represent more than one repeating cycle"],"rationale":"Follows story-map order across Judge rates using their exposure, Review a proposed regime transition."}]},"presentation":{"outcome":"Future - Business-aware expectations helps teams move from Represent more than one repeating cycle to Review a proposed regime transition across 1 backbone activity.","narrativePath":[{"backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","capabilityIds":["analysis/expectations/model-multiple-seasonal-cycles"],"health":"planned"},{"backbone":"Make expectations fit the process","step":"Judge rates using their exposure","capabilityIds":["analysis/rates/assess-bounded-count-rates"],"health":"planned"},{"backbone":"Make expectations fit the process","step":"Review a proposed regime transition","capabilityIds":["investigation/regimes/propose-reviewed-baseline-changes"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Calendar and data readiness","capabilities":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","status":"planned","area":"data","path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","health":"planned","coverage":"uncovered","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","order":10}},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","status":"planned","area":"data","path":".capabilities/data/quality/report-source-freshness.capability.yaml","health":"planned","coverage":"uncovered","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","order":20}}],"deliveryStrategy":{"release":"Future - Calendar and data readiness","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/calendar/aggregate-local-calendar-periods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/calendar/aggregate-local-calendar-periods has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Monitor the periods my business uses > Check whether the sources are current"],"rationale":"Starts with the earliest mapped step in Monitor the periods my business uses."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["data/quality/report-source-freshness"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"data/quality/report-source-freshness has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Monitor the periods my business uses > Declare local calendar boundaries"],"rationale":"Follows story-map order across Check whether the sources are current."}]},"presentation":{"outcome":"Future - Calendar and data readiness helps teams move from Declare local calendar boundaries to Check whether the sources are current across 1 backbone activity.","narrativePath":[{"backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","capabilityIds":["data/calendar/aggregate-local-calendar-periods"],"health":"planned"},{"backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","capabilityIds":["data/quality/report-source-freshness"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Cohort investigation","capabilities":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","status":"planned","area":"data","path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","health":"planned","coverage":"uncovered","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","order":10}},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","status":"planned","area":"investigation","path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","order":20}},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","status":"planned","area":"relationships","path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","order":30}}],"deliveryStrategy":{"release":"Future - Cohort investigation","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/cohorts/analyze-declared-entity-batches"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/cohorts/analyze-declared-entity-batches has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Investigate changes across populations"],"missingBackbones":[],"stepCoverageGaps":["Investigate changes across populations > Locate contributors to a total change","Investigate changes across populations > Separate performance changes from population mix"],"rationale":"Starts with the earliest mapped step in Investigate changes across populations."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["investigation/cohorts/explain-additive-segment-contributions","relationships/cohorts/separate-rate-and-population-mix"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"investigation/cohorts/explain-additive-segment-contributions has verification gaps; relationships/cohorts/separate-rate-and-population-mix has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Investigate changes across populations"],"missingBackbones":[],"stepCoverageGaps":["Investigate changes across populations > Keep each entity's evidence separate"],"rationale":"Follows story-map order across Locate contributors to a total change, Separate performance changes from population mix."}]},"presentation":{"outcome":"Future - Cohort investigation helps teams move from Keep each entity's evidence separate to Separate performance changes from population mix across 1 backbone activity.","narrativePath":[{"backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","capabilityIds":["data/cohorts/analyze-declared-entity-batches"],"health":"planned"},{"backbone":"Investigate changes across populations","step":"Locate contributors to a total change","capabilityIds":["investigation/cohorts/explain-additive-segment-contributions"],"health":"planned"},{"backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","capabilityIds":["relationships/cohorts/separate-rate-and-population-mix"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Continuous analysis","capabilities":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","status":"planned","area":"data","path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","order":30}},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","status":"planned","area":"execution","path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","order":20}},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","status":"planned","area":"execution","path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","order":10}}],"deliveryStrategy":{"release":"Future - Continuous analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["execution/state/save-and-restore-analysis-state"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"execution/state/save-and-restore-analysis-state has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Keep evidence current as observations arrive"],"missingBackbones":[],"stepCoverageGaps":["Keep evidence current as observations arrive > Analyze the next finalized samples","Keep evidence current as observations arrive > Reconcile revisions without losing history"],"rationale":"Starts with the earliest mapped step in Keep evidence current as observations arrive."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["execution/incremental/append-finalized-observations","data/revisions/reconcile-late-and-corrected-observations"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"data/revisions/reconcile-late-and-corrected-observations has verification gaps; execution/incremental/append-finalized-observations has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Keep evidence current as observations arrive"],"missingBackbones":[],"stepCoverageGaps":["Keep evidence current as observations arrive > Preserve state across restarts"],"rationale":"Follows story-map order across Analyze the next finalized samples, Reconcile revisions without losing history."}]},"presentation":{"outcome":"Future - Continuous analysis helps teams move from Preserve state across restarts to Reconcile revisions without losing history across 1 backbone activity.","narrativePath":[{"backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","capabilityIds":["execution/state/save-and-restore-analysis-state"],"health":"planned"},{"backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","capabilityIds":["execution/incremental/append-finalized-observations"],"health":"planned"},{"backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","capabilityIds":["data/revisions/reconcile-late-and-corrected-observations"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Historical evaluation","capabilities":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","status":"planned","area":"evaluation","path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","order":30}},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","status":"planned","area":"evaluation","path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","health":"planned","coverage":"uncovered","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","order":20}},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","status":"planned","area":"evaluation","path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","order":10}}],"deliveryStrategy":{"release":"Future - Historical evaluation","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["evaluation/replay/replay-historical-observations"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"evaluation/replay/replay-historical-observations has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Evaluate a monitor before relying on it"],"missingBackbones":[],"stepCoverageGaps":["Evaluate a monitor before relying on it > Compare detection with reviewed outcomes","Evaluate a monitor before relying on it > Review the tradeoff before adopting settings"],"rationale":"Starts with the earliest mapped step in Evaluate a monitor before relying on it."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["evaluation/outcomes/compare-findings-with-reviewed-incidents","evaluation/comparisons/compare-monitor-configurations"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"evaluation/comparisons/compare-monitor-configurations has verification gaps; evaluation/outcomes/compare-findings-with-reviewed-incidents has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Evaluate a monitor before relying on it"],"missingBackbones":[],"stepCoverageGaps":["Evaluate a monitor before relying on it > Reconstruct the evidence available at the time"],"rationale":"Follows story-map order across Compare detection with reviewed outcomes, Review the tradeoff before adopting settings."}]},"presentation":{"outcome":"Future - Historical evaluation helps teams move from Reconstruct the evidence available at the time to Review the tradeoff before adopting settings across 1 backbone activity.","narrativePath":[{"backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","capabilityIds":["evaluation/replay/replay-historical-observations"],"health":"planned"},{"backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","capabilityIds":["evaluation/outcomes/compare-findings-with-reviewed-incidents"],"health":"planned"},{"backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","capabilityIds":["evaluation/comparisons/compare-monitor-configurations"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Investigation memory","capabilities":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","status":"planned","area":"interfaces","path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","order":20}},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","order":10}},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","health":"planned","coverage":"uncovered","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","order":30}}],"deliveryStrategy":{"release":"Future - Investigation memory","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["investigation/history/record-reviewed-case-outcomes"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"investigation/history/record-reviewed-case-outcomes has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Reuse what an investigation teaches us"],"missingBackbones":[],"stepCoverageGaps":["Reuse what an investigation teaches us > Inspect comparable reviewed history","Reuse what an investigation teaches us > Preserve a portable evidence record"],"rationale":"Starts with the earliest mapped step in Reuse what an investigation teaches us."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["interfaces/export/package-investigation-evidence","investigation/history/retrieve-comparable-reviewed-cases"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"interfaces/export/package-investigation-evidence has verification gaps; investigation/history/retrieve-comparable-reviewed-cases has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Reuse what an investigation teaches us"],"missingBackbones":[],"stepCoverageGaps":["Reuse what an investigation teaches us > Record the reviewed outcome"],"rationale":"Follows story-map order across Inspect comparable reviewed history, Preserve a portable evidence record."}]},"presentation":{"outcome":"Future - Investigation memory helps teams move from Record the reviewed outcome to Inspect comparable reviewed history across 1 backbone activity.","narrativePath":[{"backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","capabilityIds":["investigation/history/record-reviewed-case-outcomes"],"health":"planned"},{"backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","capabilityIds":["interfaces/export/package-investigation-evidence"],"health":"planned"},{"backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","capabilityIds":["investigation/history/retrieve-comparable-reviewed-cases"],"health":"planned"}],"coverageSignals":[]}},{"release":"Incomplete-period monitoring","capabilities":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","status":"implemented","area":"analysis","path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","health":"review","coverage":"partial","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","order":20}},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","status":"implemented","area":"data","path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","health":"review","coverage":"partial","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","references":{"total":10,"readable":10,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":13,"excerpt":"\"seasonal-residual-v1\", \"adaptive-seasonal-v1\", \"multi-resolution-v1\","},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":297,"excerpt":"if (self.data_policy or self.coverage or self.available_at or self.revisions) and self.timestamps is None:"},{"reference":"src/anomalyzer/contracts.py","line":298,"excerpt":"raise ValueError(\"source policies/coverage/revisions require original timestamps\")"},{"reference":"src/anomalyzer/contracts.py","line":437,"excerpt":"total_points = sum(len(dataset.values) + len(dataset.revisions) for dataset in self.datasets)"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":64,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse.py","line":66,"excerpt":"if coverage != \"complete\" or previous[\"coverage\"] != \"complete\" or ds.available_at or ds.revisions:"},{"reference":"src/anomalyzer/sparse.py","line":83,"excerpt":"observed_at = timestamp(revision.timestamp, ds.timezone)"},{"reference":"src/anomalyzer/sparse_views.py","line":1,"excerpt":"\"\"\"Schema-1.1 native and complete-day evidence with independent view eligibility.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":28,"excerpt":"case.revision_id = f\"{resolution}:{case.revision_id}\""},{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/orchestrator.py","line":106,"excerpt":"if request.config.recipe == \"multi-resolution-v1\":"},{"reference":"src/anomalyzer/orchestrator.py","line":107,"excerpt":"from .sparse_views import analyze_multi_request"},{"reference":"src/anomalyzer/orchestrator.py","line":133,"excerpt":"if \"data_policy\" in source.quality and request.config.recipe != \"multi-resolution-v1\""},{"reference":"src/anomalyzer/orchestrator.py","line":166,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":180,"excerpt":"reset_positions=active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of),"},{"reference":"src/anomalyzer/relationships.py","line":160,"excerpt":"source_timestamps = {source.dataset.id: source.times[source_index]"},{"reference":"src/anomalyzer/relationships.py","line":162,"excerpt":"cutoff = max(source_timestamps.values())"},{"reference":"src/anomalyzer/relationships.py","line":166,"excerpt":"source_indexes=source_indexes, source_timestamps=source_timestamps,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":124,"excerpt":"def test_sparse_phase_without_support_is_explicitly_unavailable():"},{"reference":"tests/test_sparse.py","line":181,"excerpt":"ds[\"timestamps\"][1] = \"2026-02-01T00:00:00Z\""},{"reference":"docs/sparse-data.md","line":30,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":87,"excerpt":"otherwise bounds come from the supplied as-of-available timestamp range."},{"reference":"docs/sparse-data.md","line":105,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"examples/sparse_examples.py","line":59,"excerpt":"daily = request(dataset([1.0]*60, data_policy={\"measurement\":\"additive_volume\"}), recipe=\"multi-resolution-v1\","},{"reference":"examples/sparse_examples.py","line":95,"excerpt":"daily = next(item for item in outputs[\"unfinished_daily\"].dataset_results if item.resolution == \"completed_period\")"},{"reference":"examples/render_readme_progression.py","line":686,"excerpt":"def figure_multi_resolution():"},{"reference":"examples/render_readme_progression.py","line":699,"excerpt":"\"config\": {\"recipe\": \"multi-resolution-v1\", \"aggregate_function\": \"sum\"},"},{"reference":"examples/render_readme_progression.py","line":710,"excerpt":"assert partial[\"included_in_completed_period_view\"] is False"},{"reference":"examples/render_readme_progression.py","line":767,"excerpt":"complete_count = result.data_quality[\"views\"][\"completed_period\"][\"observation_count\"]"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":51,"excerpt":"def valid_reset_points(cls, values):"},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":71,"excerpt":"def numeric_values(cls, data):"},{"reference":"src/anomalyzer/contracts.py","line":85,"excerpt":"if self.period_statuses is not None and len(self.period_statuses) != len(self.values):"},{"reference":"src/anomalyzer/contracts.py","line":86,"excerpt":"raise ValueError(\"period_statuses and values must have equal length\")"},{"reference":"src/anomalyzer/sparse.py","line":3,"excerpt":"Calendar coordinates remain dense only inside an explicitly bounded input grid."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":42,"excerpt":"\"\"\"Return the explicit grid, eligible model values and inspectable provenance.\"\"\""},{"reference":"src/anomalyzer/sparse.py","line":48,"excerpt":"statuses = ds.period_statuses or [\"complete\"] * len(ds.values)"},{"reference":"src/anomalyzer/sparse.py","line":54,"excerpt":"\"complete\" if value is not None else \"empty\" if policy.absence_means_zero else \"unknown\")"},{"reference":"src/anomalyzer/sparse_views.py","line":72,"excerpt":"measurements = quality[\"rule_values\"]"},{"reference":"src/anomalyzer/sparse_views.py","line":75,"excerpt":"if policy.observation_basis != \"calendar\":"},{"reference":"src/anomalyzer/sparse_views.py","line":82,"excerpt":"if policy.measurement != \"additive_volume\" or cfg.aggregate_function != \"sum\":"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":98,"excerpt":"daily_times, daily_values, coverage, records, available_at = [], [], [], [], []"},{"reference":"src/anomalyzer/orchestrator.py","line":28,"excerpt":"if not any(row.get(\"available_at\") and (row.get(\"value\") is not None or row.get(\"coverage\") == \"excluded\") for row in samples):"},{"reference":"src/anomalyzer/orchestrator.py","line":63,"excerpt":"unavailable values. Dropping only that leading run lets the ordinary engine"},{"reference":"src/anomalyzer/orchestrator.py","line":127,"excerpt":"prepared_points += len(source.values) + len(dataset.revisions)"},{"reference":"src/anomalyzer/orchestrator.py","line":139,"excerpt":"source.quality.get(\"rule_values\", source.values)[:source.quality.get(\"grid_count\", source.quality[\"observation_count\"])], dataset.rules,"},{"reference":"src/anomalyzer/orchestrator.py","line":191,"excerpt":"[dict(value=p.value, available_at=p.latest_source_cutoff if any(p.source_available_at.values()) else None) for p in lineage])"},{"reference":"src/anomalyzer/relationships.py","line":36,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/relationships.py","line":58,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":98,"excerpt":"value, unavailable = None, \"zero_denominator\""},{"reference":"src/anomalyzer/relationships.py","line":131,"excerpt":"y, x = response.values[index], predictor.values[predictor_index]"},{"reference":"src/anomalyzer/relationships.py","line":141,"excerpt":"source_values = [source.values[source_index]"},{"reference":"src/anomalyzer/early.py","line":58,"excerpt":"for index, (observed_at, value) in enumerate(zip(times, values)):"},{"reference":"src/anomalyzer/early.py","line":137,"excerpt":"criterion_met and policy.allow_explicit_rules)"},{"reference":"tests/test_sparse.py","line":45,"excerpt":"def test_zero_is_only_authorized_for_complete_empty_additive_volume():"},{"reference":"tests/test_sparse.py","line":46,"excerpt":"request = body([2, None, None, 0, 9], data_policy={\"missing\": \"zero_for_empty\", \"measurement\": \"additive_volume\"},"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":52,"excerpt":"assert rows[1][\"reason\"] == \"confirmed_empty_additive_volume\""},{"reference":"tests/test_sparse.py","line":58,"excerpt":"def test_global_collection_contract_authorizes_omitted_slots_but_not_explicit_unknown():"},{"reference":"docs/sparse-data.md","line":4,"excerpt":"Schema 1.1 therefore preserves missing values by default. An expectation can be"},{"reference":"docs/sparse-data.md","line":33,"excerpt":"data_policy=DataPolicy(),  # preserve missing; hour 2 is also unknown"},{"reference":"docs/sparse-data.md","line":41,"excerpt":"| Numeric value, including zero, `coverage=\"complete\"` | A measurement eligible for rules and model support. Explicit zero remains zero. |"},{"reference":"docs/sparse-data.md","line":43,"excerpt":"| `coverage=\"empty\"` | Caller confirms a finalized period with no reported volume. Preserve as missing unless an additive zero policy explicitly authorizes conversion. |"},{"reference":"docs/sparse-data.md","line":44,"excerpt":"| `coverage=\"unknown\"` | Collection/exposure is unknown; never filled even under a global empty-volume contract. |"},{"reference":"examples/sparse_examples.py","line":22,"excerpt":"preserve = request(dataset(values, coverage=[\"empty\" if v is None else \"complete\" for v in values],"},{"reference":"examples/sparse_examples.py","line":23,"excerpt":"data_policy={\"measurement\": \"additive_volume\"}))"},{"reference":"examples/sparse_examples.py","line":25,"excerpt":"zero[\"datasets\"][0][\"data_policy\"][\"missing\"] = \"zero_for_empty\""},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":28,"excerpt":"coverage=[\"complete\"]*2+[\"unknown\"]+[\"complete\"]*6+[\"unknown\"]+[\"complete\"]*3+[\"incomplete\"], data_policy={}))"},{"reference":"examples/render_readme_progression.py","line":135,"excerpt":"def chart(svg, values, overlays, x, y, width, height, *, y_bounds=None,"},{"reference":"examples/render_readme_progression.py","line":139,"excerpt":"all_values = [values] + [series for _, series, *_ in overlays]"},{"reference":"examples/render_readme_progression.py","line":140,"excerpt":"low, high = y_bounds or bounds(*all_values)"},{"reference":"examples/render_readme_progression.py","line":166,"excerpt":"if value is not None and (index == 0 or values[index-1] is None) and (index == count-1 or values[index+1] is None):"},{"reference":"examples/render_readme_progression.py","line":206,"excerpt":"values = [80 + 0.65 * index + [0, 0.25, -0.15, 0.10, -0.20][index % 5]"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":242,"excerpt":"missing: Literal[\"preserve\", \"zero_for_empty\"] = \"preserve\""},{"reference":"src/anomalyzer/contracts.py","line":245,"excerpt":"observation_basis: Literal[\"calendar\", \"event_index\"] = \"calendar\""},{"reference":"src/anomalyzer/contracts.py","line":258,"excerpt":"if self.observation_basis == \"event_index\" and ("},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":49,"excerpt":"for index, (text, value, status) in enumerate(zip(ds.timestamps, ds.values, statuses)):"},{"reference":"src/anomalyzer/sparse.py","line":57,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":63,"excerpt":"if cfg.duplicate_policy == \"error\" or policy.observation_basis == \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":64,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":128,"excerpt":"policy_zero_children=sum(quality[\"samples\"][i][\"reason\"] == \"confirmed_empty_additive_volume\" for i in indexes),"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":134,"excerpt":"timestamps=daily_times, values=daily_values, frequency=\"1d\", coverage=coverage, available_at=available_at,"},{"reference":"src/anomalyzer/orchestrator.py","line":37,"excerpt":"assessment.baseline[\"known_input_availability_cutoff\"] = cutoffs[assessment.index]"},{"reference":"src/anomalyzer/orchestrator.py","line":38,"excerpt":"assessment.assumptions.append(\"Availability cutoff is conservative across known prior inputs; unknown arrival history cannot establish delivery-time detection performance.\")"},{"reference":"src/anomalyzer/orchestrator.py","line":166,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":178,"excerpt":"relationship.id, derived.timestamps or [], derived.values,"},{"reference":"src/anomalyzer/orchestrator.py","line":180,"excerpt":"reset_positions=active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of),"},{"reference":"src/anomalyzer/relationships.py","line":40,"excerpt":"raise ValueError(\"relationship timestamps cannot be positional\")"},{"reference":"src/anomalyzer/relationships.py","line":57,"excerpt":"for index in range(relationship.lag, coefficient_cutoff + 1):"},{"reference":"src/anomalyzer/relationships.py","line":58,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":96,"excerpt":"if relationship.zero_denominator == \"error\":"},{"reference":"src/anomalyzer/relationships.py","line":97,"excerpt":"raise ValueError(f\"zero denominator at aligned index {index}\")"},{"reference":"src/anomalyzer/early.py","line":140,"excerpt":"unavailable_rules = [dict(id=rule.id, reason=\"no_comparable_prior\" if previous is None else \"relative_change_from_zero\")"},{"reference":"src/anomalyzer/early.py","line":147,"excerpt":"target_id=target_id, index=index, timestamp=observed_at,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":68,"excerpt":"{\"observation_basis\": \"event_index\", \"grid_start\": \"2026-01-01T00:00:00Z\"},"},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":129,"excerpt":"assert any(row[\"reason\"] == \"no_training_support_for_calendar_phase\" for row in result.methods[0].diagnostics[\"unavailable\"])"},{"reference":"docs/sparse-data.md","line":30,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":105,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"docs/sparse-data.md","line":110,"excerpt":"rejected. Event-index timestamps must be unique; ties are refused rather than"},{"reference":"docs/sparse-data.md","line":141,"excerpt":"original calendar indexes and timestamps. Post-gap learning requires an explicit"},{"reference":"docs/sparse-data.md","line":155,"excerpt":"## Comparable event-index observations"},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":38,"excerpt":"events[\"timestamps\"] = [(datetime(2026,1,1,tzinfo=timezone.utc)+timedelta(hours=h)).isoformat() for h in hours]"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":81,"excerpt":"preserve, zero = [outputs[n].dataset_results[0] for n in (\"preserve\", \"zero_empty\")]"},{"reference":"examples/render_readme_progression.py","line":654,"excerpt":"svg.text(430, 467, \"Sample index (zero-based)\", 15, MUTED, anchor=\"middle\")"},{"reference":"examples/render_readme_progression.py","line":697,"excerpt":"\"timestamps\": [(origin + timedelta(hours=index)).isoformat()"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"},{"reference":"examples/render_readme_progression.py","line":887,"excerpt":"shared_high = max(v for v in observed(\"preserve\")+expectation(\"preserve\")+observed(\"zero_empty\")+expectation(\"zero_empty\") if v is not None)*1.12"},{"reference":"examples/render_readme_progression.py","line":897,"excerpt":"header(svg, \"15\", \"Keep zero, unknown and unfinished distinct\", \"Blue zeros are actual measurements; gaps stay gaps; the orange open-period value is excluded.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","order":30}},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","status":"implemented","area":"data","path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","health":"review","coverage":"partial","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","order":10}}],"deliveryStrategy":{"release":"Incomplete-period monitoring","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/quality/exclude-incomplete-periods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/quality/exclude-incomplete-periods needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Monitor while periods are still open"],"missingBackbones":[],"stepCoverageGaps":["Monitor while periods are still open > Assess sparse source data","Monitor while periods are still open > Compare finalized subperiods"],"rationale":"Starts with the earliest mapped step in Monitor while periods are still open."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/resolution/monitor-within-unfinished-days","data/quality/analyze-sparse-observations"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/resolution/monitor-within-unfinished-days needs review; data/quality/analyze-sparse-observations needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Monitor while periods are still open"],"missingBackbones":[],"stepCoverageGaps":["Monitor while periods are still open > Keep unfinished data visible"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/resolution/monitor-within-unfinished-days, data/quality/analyze-sparse-observations."}]},"presentation":{"outcome":"Incomplete-period monitoring helps teams move from Keep unfinished data visible to Assess sparse source data across 1 backbone activity.","narrativePath":[{"backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","capabilityIds":["data/quality/exclude-incomplete-periods"],"health":"review"},{"backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","capabilityIds":["analysis/resolution/monitor-within-unfinished-days"],"health":"review"},{"backbone":"Monitor while periods are still open","step":"Assess sparse source data","capabilityIds":["data/quality/analyze-sparse-observations"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Monitor while periods are still open > Assess sparse source data","message":"Monitor while periods are still open > Assess sparse source data has needs-review coverage."},{"kind":"weak","label":"Monitor while periods are still open > Compare finalized subperiods","message":"Monitor while periods are still open > Compare finalized subperiods has needs-review coverage."},{"kind":"weak","label":"Monitor while periods are still open > Keep unfinished data visible","message":"Monitor while periods are still open > Keep unfinished data visible has needs-review coverage."}]}},{"release":"Python integration","capabilities":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","health":"review","coverage":"partial","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","order":10}},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","health":"action","coverage":"partial","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","references":{"total":11,"readable":11,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[{"code":"declared-gap","message":"interfaces/python/install-versioned-library: First immutable GitHub Release requires owner setup and a reviewed version tag.","capabilityId":"interfaces/python/install-versioned-library","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\python\\install-versioned-library.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Resolve or explicitly ignore verification gaps.","topFindings":[{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/publish_release.py","line":43,"excerpt":"\"Linux/Python 3.12 Lambda consumer verified offline. \""},{"reference":"scripts/verify_distribution.py","line":41,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":6,"excerpt":"bundle=\"$root/.tmp/lambda-consumer\""},{"reference":"scripts/verify_lambda.sh","line":18,"excerpt":"python -m pip download --only-binary=:all: --require-hashes \\"},{"reference":"scripts/verify_lambda.sh","line":20,"excerpt":"python -m pip install --no-index --only-binary=:all: --require-hashes \\"},{"reference":".github/workflows/ci.yml","line":29,"excerpt":"- run: uv run python examples/generate.py"},{"reference":".github/workflows/ci.yml","line":32,"excerpt":"- run: uv run python examples/run_demo.py"},{"reference":".github/workflows/ci.yml","line":33,"excerpt":"- run: uv run python examples/run_agentic_demo.py"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/ci.yml","line":54,"excerpt":"run: uv export --locked --no-dev --no-emit-project --output-file dist/runtime-dependencies.txt"},{"reference":"docs/releases/README.md","line":12,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":15,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":58,"excerpt":"uv run python examples/run_demo.py"},{"reference":"docs/releases/README.md","line":59,"excerpt":"uv run python examples/run_agentic_demo.py"},{"reference":"docs/releases/README.md","line":75,"excerpt":"versions, and tests a bundled Linux x86_64/Python 3.12 consumer in AWS's Lambda"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","evidence":[{"reference":"scripts/verify_distribution.py","line":1,"excerpt":"\"\"\"Install wheel AND sdist in clean environments outside the source tree (requires uv).\"\"\""},{"reference":"scripts/verify_distribution.py","line":42,"excerpt":"run([\"uv\", \"pip\", \"install\", \"--python\", python, \"--no-deps\", artifact], outside, env=env)"},{"reference":"examples/library/guard_smoke.py","line":32,"excerpt":"analyze_relationships, get_case, replay_policy,"},{"reference":"examples/library/guard_smoke.py","line":49,"excerpt":"assert legacy.methods[0].diagnostics[\"detection_readiness\"]"},{"reference":"examples/library/guard_smoke.py","line":65,"excerpt":"\"relationships\": [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"cost\","},{"reference":"examples/library/guard_smoke.py","line":69,"excerpt":"result = analyze_relationships(request)"},{"reference":"examples/library/guard_smoke.py","line":76,"excerpt":"short = analyze_relationships({**request.model_dump(mode=\"json\"),"},{"reference":".github/workflows/ci.yml","line":59,"excerpt":"- name: Test built wheel and sdist outside checkout"},{"reference":"docs/releases/README.md","line":13,"excerpt":"| Source distribution (`.tar.gz`, or sdist) | Source needed to build the package. Useful for rebuilding and inspection; installation also needs the declared build backend. |"},{"reference":"docs/releases/README.md","line":208,"excerpt":"`analyze_series`, `analyze_relationships`, `get_case`, `replay_policy` (also"},{"reference":"docs/releases/README.md","line":211,"excerpt":"`anomalyzer.contracts`. See the self-contained"},{"reference":"docs/releases/README.md","line":213,"excerpt":"all APIs, recipes, point/pattern evidence, relationships, cases and policy"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":34,"excerpt":"return hashlib.sha256(path.read_bytes()).hexdigest()"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/release.py","line":85,"excerpt":"suffix = f\" --hash=sha256:{digest(wheel)}\\n\""},{"reference":"scripts/release.py","line":92,"excerpt":"assets = sorted(p for p in dist.iterdir() if p.is_file()"},{"reference":"scripts/publish_release.py","line":1,"excerpt":"\"\"\"Publish tested assets without replacing tags or existing assets (GitHub CLI auth).\"\"\""},{"reference":"scripts/publish_release.py","line":25,"excerpt":"remote_sha = json.loads(gh(\"api\", f\"repos/{REPOSITORY}/commits/{tag}\"))[\"sha\"]"},{"reference":"scripts/publish_release.py","line":26,"excerpt":"if remote_sha != revision:"},{"reference":"scripts/publish_release.py","line":27,"excerpt":"raise ValueError(\"Remote tag no longer points to the tested source revision\")"},{"reference":"scripts/publish_release.py","line":28,"excerpt":"endpoint = f\"repos/{REPOSITORY}/releases/tags/{tag}\""},{"reference":"scripts/verify_distribution.py","line":41,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":25,"excerpt":"sha256sum *.whl > /output/wheelhouse-SHA256SUMS"},{"reference":"examples/library/guard_smoke.py","line":39,"excerpt":"assert anomalyzer.__version__ == importlib.metadata.version(\"anomalyzer\")"},{"reference":".github/workflows/ci.yml","line":44,"excerpt":"- name: Check release tag before building"},{"reference":".github/workflows/ci.yml","line":47,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/ci.yml","line":48,"excerpt":"run: python scripts/release.py check-version --tag \"$RELEASE_TAG\""},{"reference":".github/workflows/ci.yml","line":51,"excerpt":"- name: Check package metadata"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/release.yml","line":31,"excerpt":"name: release-dist-${{ github.sha }}"},{"reference":".github/workflows/release.yml","line":33,"excerpt":"- name: Stage tested assets for owner publication"},{"reference":".github/workflows/release.yml","line":36,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/release.yml","line":37,"excerpt":"run: python scripts/publish_release.py --tag \"$RELEASE_TAG\" --revision \"$GITHUB_SHA\""},{"reference":"docs/releases/README.md","line":3,"excerpt":"Anomalyzer is a normal Python dependency. Version **0.2.0** is the first release"},{"reference":"docs/releases/README.md","line":12,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":14,"excerpt":"| Version tag (`v0.2.0`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |"},{"reference":"docs/releases/README.md","line":15,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"it for package metadata and the public `__version__` imports it."},{"reference":"tests/test_release.py","line":1,"excerpt":"\"\"\"Release safety: reject wrong tags and refuse changes to published artifacts.\"\"\""},{"reference":"tests/test_release.py","line":21,"excerpt":"def test_release_refuses_mismatched_or_unsupported_tag(tag):"},{"reference":"tests/test_release.py","line":23,"excerpt":"release.check_tag(tag)"},{"reference":"tests/test_release.py","line":52,"excerpt":"def test_published_matching_release_is_noop(monkeypatch, tmp_path):"},{"reference":"tests/test_release.py","line":81,"excerpt":"\"version\": release.version(), \"source_revision\": \"a\" * 40, \"tag\": f\"v{release.version()}\","}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","order":20}}],"deliveryStrategy":{"release":"Python integration","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/python/analyze-through-public-api"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/python/analyze-through-public-api needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Embed analysis in my application"],"missingBackbones":[],"stepCoverageGaps":["Embed analysis in my application > Install a released dependency"],"rationale":"Starts with the earliest mapped step in Embed analysis in my application."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/python/install-versioned-library"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/python/install-versioned-library has verification gaps; interfaces/python/install-versioned-library needs action","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Embed analysis in my application"],"missingBackbones":[],"stepCoverageGaps":["Embed analysis in my application > Call the public API"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/python/install-versioned-library."}]},"presentation":{"outcome":"Python integration helps teams move from Call the public API to Install a released dependency across 1 backbone activity.","narrativePath":[{"backbone":"Embed analysis in my application","step":"Call the public API","capabilityIds":["interfaces/python/analyze-through-public-api"],"health":"review"},{"backbone":"Embed analysis in my application","step":"Install a released dependency","capabilityIds":["interfaces/python/install-versioned-library"],"health":"action"}],"coverageSignals":[{"kind":"weak","label":"Embed analysis in my application > Call the public API","message":"Embed analysis in my application > Call the public API has needs-review coverage."},{"kind":"weak","label":"Embed analysis in my application > Install a released dependency","message":"Embed analysis in my application > Install a released dependency has needs-action coverage."}]}},{"release":"Read-only MCP server","capabilities":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","health":"review","coverage":"partial","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Analysis tool calls return structured content together with compact text.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","order":10}},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","order":20}}],"deliveryStrategy":{"release":"Read-only MCP server","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/mcp/connect-read-only-local-server"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/mcp/connect-read-only-local-server needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Give an agent access to local evidence"],"missingBackbones":[],"stepCoverageGaps":["Give an agent access to local evidence > Drill into findings and evidence"],"rationale":"Starts with the earliest mapped step in Give an agent access to local evidence."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/mcp/investigate-findings-and-evidence"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/mcp/investigate-findings-and-evidence needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Give an agent access to local evidence"],"missingBackbones":[],"stepCoverageGaps":["Give an agent access to local evidence > Connect a local server"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/mcp/investigate-findings-and-evidence."}]},"presentation":{"outcome":"Read-only MCP server helps teams move from Connect a local server to Drill into findings and evidence across 1 backbone activity.","narrativePath":[{"backbone":"Give an agent access to local evidence","step":"Connect a local server","capabilityIds":["interfaces/mcp/connect-read-only-local-server"],"health":"review"},{"backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","capabilityIds":["interfaces/mcp/investigate-findings-and-evidence"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Give an agent access to local evidence > Connect a local server","message":"Give an agent access to local evidence > Connect a local server has needs-review coverage."},{"kind":"weak","label":"Give an agent access to local evidence > Drill into findings and evidence","message":"Give an agent access to local evidence > Drill into findings and evidence has needs-review coverage."}]}},{"release":"Related-metric analysis","capabilities":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","health":"review","coverage":"partial","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","order":20}},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","health":"review","coverage":"partial","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","order":30}},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","status":"implemented","area":"relationships","path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","health":"review","coverage":"partial","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Difference relationships require identical declared source units and matching output units.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","order":10}},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","status":"implemented","area":"relationships","path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","health":"review","coverage":"partial","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":22,"excerpt":"cases = compose_cases([d for d in datasets if d.resolution == resolution],"},{"reference":"src/anomalyzer/sparse_views.py","line":146,"excerpt":"native_sources = [d.model_copy(update={\"data_policy\": d.data_policy or DataPolicy()}) for d in request.datasets]"},{"reference":"src/anomalyzer/sparse_views.py","line":149,"excerpt":"native_request = request.model_copy(update={\"datasets\": native_sources, \"config\": native_config})"},{"reference":"src/anomalyzer/sparse_views.py","line":169,"excerpt":"daily_request = request.model_copy(update={\"datasets\": sources, \"config\": daily_config, \"relationships\": daily_relationships})"},{"reference":"src/anomalyzer/agent.py","line":14,"excerpt":"\"\"\"Use legacy semantics for a Dataset; explicit V11 fields select source policies.\"\"\""},{"reference":"src/anomalyzer/agent.py","line":20,"excerpt":"name in dataset for name in (\"data_policy\", \"coverage\", \"available_at\", \"revisions\", \"rules\", \"completed_period_rules\")):"},{"reference":"src/anomalyzer/agent.py","line":23,"excerpt":"return analyze({\"schema_version\": \"1.1\", \"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"src/anomalyzer/agent.py","line":29,"excerpt":"return analyze({\"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"tests/test_sparse.py","line":18,"excerpt":"return {\"schema_version\": \"1.1\", \"datasets\": [{\"id\": \"series\", \"entity\": {\"customer\": \"test\"},"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":88,"excerpt":"def test_declared_duplicate_aggregation_preserves_provenance(policy, value):"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":118,"excerpt":"changed[\"datasets\"][0][\"values\"][-10:] = [999]*10"},{"reference":"docs/sparse-data.md","line":35,"excerpt":"result = analyze({\"schema_version\": \"1.1\", \"datasets\": [source.model_dump(mode=\"json\")]})"},{"reference":"docs/sparse-data.md","line":136,"excerpt":"incomplete, excluded or reset boundaries, regardless of point-reference policy."},{"reference":"docs/sparse-data.md","line":142,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":172,"excerpt":"basis. A caller-declared maximum time separation can limit change comparisons."},{"reference":"docs/sparse-data.md","line":218,"excerpt":"Daily lagged, normalized-residual and joint-condition definitions remain"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":70,"excerpt":"\"\"\"Sum only wholly observed finalized additive children. Never average ratios.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":175,"excerpt":"applicability=\"completed-day view supports ratio-of-sums and differences of compatible additive sums only\","},{"reference":"src/anomalyzer/relationships.py","line":90,"excerpt":"used = [(numerator, index), (denominator, index)]"},{"reference":"src/anomalyzer/relationships.py","line":173,"excerpt":"\"complete\" if source.values[i] is not None else \"unknown_or_incomplete\" for source, i in used}"},{"reference":"tests/test_sparse.py","line":231,"excerpt":"request[\"relationships\"] = [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"series\", \"denominator\": \"calls\", \"units\": \"dollars_per_call\"}]"},{"reference":"tests/test_sparse.py","line":260,"excerpt":"def test_daily_rules_are_not_silently_reused_from_hourly_rules():"},{"reference":"tests/test_sparse.py","line":279,"excerpt":"def test_adaptive_gap_does_not_silently_train_a_post_gap_episode():"},{"reference":"tests/test_sparse.py","line":322,"excerpt":"def test_restart_policy_cannot_reuse_a_prior_regime_reference():"},{"reference":"tests/test_sparse.py","line":431,"excerpt":"assert row[\"complete\"] and row[\"aggregate_value\"] == 24 and not row[\"model_eligible\"]"},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":120,"excerpt":"It then collects `calibration_size` eligible forecast residuals, not elapsed slots."},{"reference":"docs/sparse-data.md","line":142,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":213,"excerpt":"partial sums are progress metadata, never a completed total. Empty days are not"},{"reference":"docs/sparse-data.md","line":216,"excerpt":"Daily ratios divide summed compatible numerator and denominator quantities;"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":130,"excerpt":"source_grid_indexes=indexes, observed_sum=(math.fsum(measurements[i] for i in indexes if measurements[i] is not None)"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":36,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/agent.py","line":34,"excerpt":"def analyze_relationships(request: RequestV11 | dict) -> ResultV11:"},{"reference":"tests/test_sparse.py","line":10,"excerpt":"from anomalyzer import analyze, analyze_series, analyze_relationships, get_case, replay_policy"},{"reference":"tests/test_sparse.py","line":41,"excerpt":"assert right.data_quality[\"samples\"][8][\"source_index\"] == -1"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":61,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, 0, 0, 2]"},{"reference":"tests/test_sparse.py","line":63,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, None, 0, 2]"},{"reference":"docs/sparse-data.md","line":21,"excerpt":"works through `analyze`, `analyze_relationships`, CLI JSON and read-only MCP."},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":72,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":76,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":106,"excerpt":"retains its unique-source requirement. With an explicit `data_policy`, complete,"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","order":30}}],"deliveryStrategy":{"release":"Related-metric analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["relationships/quality/require-compatible-sources","relationships/resolution/compare-finalized-subperiods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"relationships/quality/require-compatible-sources needs review; relationships/resolution/compare-finalized-subperiods needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Investigate relationships I declare","Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Investigate relationships I declare > Check delayed responses","Investigate relationships I declare > Compare ratios and metric gaps"],"rationale":"Starts with one step from each mapped backbone: Investigate relationships I declare, Monitor the periods my business uses."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"relationships/comparisons/compare-declared-metrics needs review; relationships/comparisons/evaluate-delayed-response needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Investigate relationships I declare"],"missingBackbones":["Monitor the periods my business uses"],"stepCoverageGaps":["Investigate relationships I declare > Align comparable sources","Monitor the periods my business uses > Compare related metrics while the period is open"],"rationale":"Prioritizes capabilities with delivery risk signals: relationships/comparisons/compare-declared-metrics, relationships/comparisons/evaluate-delayed-response."}]},"presentation":{"outcome":"Related-metric analysis helps teams move from Align comparable sources to Compare related metrics while the period is open across 2 backbone activities.","narrativePath":[{"backbone":"Investigate relationships I declare","step":"Align comparable sources","capabilityIds":["relationships/quality/require-compatible-sources"],"health":"review"},{"backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","capabilityIds":["relationships/comparisons/compare-declared-metrics"],"health":"review"},{"backbone":"Investigate relationships I declare","step":"Check delayed responses","capabilityIds":["relationships/comparisons/evaluate-delayed-response"],"health":"review"},{"backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","capabilityIds":["relationships/resolution/compare-finalized-subperiods"],"health":"review"}],"coverageSignals":[{"kind":"missing","label":"Monitor the periods my business uses","message":"Monitor the periods my business uses is not covered in the recommended slice."},{"kind":"weak","label":"Investigate relationships I declare > Align comparable sources","message":"Investigate relationships I declare > Align comparable sources has needs-review coverage."},{"kind":"weak","label":"Investigate relationships I declare > Check delayed responses","message":"Investigate relationships I declare > Check delayed responses has needs-review coverage."},{"kind":"weak","label":"Investigate relationships I declare > Compare ratios and metric gaps","message":"Investigate relationships I declare > Compare ratios and metric gaps has needs-review coverage."},{"kind":"weak","label":"Monitor the periods my business uses > Compare related metrics while the period is open","message":"Monitor the periods my business uses > Compare related metrics while the period is open has needs-review coverage."}]}},{"release":"Trend and seasonal analysis","capabilities":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","health":"review","coverage":"partial","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Moving-range findings identify variation changes separately from location changes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","order":60}},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","health":"review","coverage":"partial","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","order":40}},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","health":"review","coverage":"partial","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","order":10}},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","health":"review","coverage":"partial","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","order":30}},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","health":"review","coverage":"partial","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","order":50}},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/select-season-length.capability.yaml","health":"review","coverage":"partial","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"An explicit season length is honored and reported as explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","order":20}}],"deliveryStrategy":{"release":"Trend and seasonal analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["analysis/expectations/model-seasonal-level-and-growth"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"analysis/expectations/model-seasonal-level-and-growth needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Refine what ordinary behavior means"],"missingBackbones":[],"stepCoverageGaps":["Refine what ordinary behavior means > Choose a repeating cycle","Refine what ordinary behavior means > Find sustained residual changes","Refine what ordinary behavior means > Follow evolving growth","Refine what ordinary behavior means > Protect later comparisons","Refine what ordinary behavior means > Restart after a reviewed change"],"rationale":"Starts with the earliest mapped step in Refine what ordinary behavior means."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/expectations/select-season-length","analysis/expectations/protect-reference-from-extremes","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/detection/identify-residual-patterns needs review; analysis/expectations/adapt-to-changing-growth needs review; analysis/expectations/protect-reference-from-extremes needs review; analysis/expectations/reset-declared-regimes needs review; analysis/expectations/select-season-length needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Refine what ordinary behavior means"],"missingBackbones":[],"stepCoverageGaps":["Refine what ordinary behavior means > Account for growth and seasonality"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/expectations/select-season-length, analysis/expectations/protect-reference-from-extremes, analysis/expectations/adapt-to-changing-growth, analysis/expectations/reset-declared-regimes, analysis/detection/identify-residual-patterns."}]},"presentation":{"outcome":"Trend and seasonal analysis helps teams move from Account for growth and seasonality to Find sustained residual changes across 1 backbone activity.","narrativePath":[{"backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","capabilityIds":["analysis/expectations/model-seasonal-level-and-growth"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","capabilityIds":["analysis/expectations/select-season-length"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","capabilityIds":["analysis/expectations/protect-reference-from-extremes"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","capabilityIds":["analysis/expectations/adapt-to-changing-growth"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","capabilityIds":["analysis/expectations/reset-declared-regimes"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","capabilityIds":["analysis/detection/identify-residual-patterns"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Refine what ordinary behavior means > Account for growth and seasonality","message":"Refine what ordinary behavior means > Account for growth and seasonality has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Choose a repeating cycle","message":"Refine what ordinary behavior means > Choose a repeating cycle has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Find sustained residual changes","message":"Refine what ordinary behavior means > Find sustained residual changes has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Follow evolving growth","message":"Refine what ordinary behavior means > Follow evolving growth has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Protect later comparisons","message":"Refine what ordinary behavior means > Protect later comparisons has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Restart after a reviewed change","message":"Refine what ordinary behavior means > Restart after a reviewed change has needs-review coverage."}]}}]},"summary":{"total":48,"ok":0,"review":30,"action":1,"planned":17}};
+const report = {"project":"Anomalyzer","capabilities":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","health":"review","coverage":"partial","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Moving-range findings identify variation changes separately from location changes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","order":60}},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","health":"review","coverage":"partial","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","order":40}},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","health":"review","coverage":"partial","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","order":20}},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","health":"review","coverage":"partial","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","order":30}},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","health":"review","coverage":"partial","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","order":40}},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","health":"review","coverage":"partial","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","order":10}},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","health":"review","coverage":"partial","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","order":30}},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","health":"review","coverage":"partial","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","order":50}},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/select-season-length.capability.yaml","health":"review","coverage":"partial","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"An explicit season length is honored and reported as explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","order":20}},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","status":"implemented","area":"analysis","path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","health":"review","coverage":"partial","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","order":20}},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","status":"implemented","area":"data","path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","health":"review","coverage":"partial","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","order":10}},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","status":"implemented","area":"data","path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","health":"review","coverage":"partial","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","references":{"total":10,"readable":10,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":13,"excerpt":"\"seasonal-residual-v1\", \"adaptive-seasonal-v1\", \"multi-resolution-v1\","},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":304,"excerpt":"if (self.data_policy or self.coverage or self.available_at or self.revisions) and self.timestamps is None:"},{"reference":"src/anomalyzer/contracts.py","line":305,"excerpt":"raise ValueError(\"source policies/coverage/revisions require original timestamps\")"},{"reference":"src/anomalyzer/contracts.py","line":310,"excerpt":"if self.timestamps is None or self.data_policy is None or self.data_policy.observation_basis != \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":67,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse.py","line":69,"excerpt":"if coverage != \"complete\" or previous[\"coverage\"] != \"complete\" or ds.available_at or ds.revisions:"},{"reference":"src/anomalyzer/sparse.py","line":86,"excerpt":"observed_at = timestamp(revision.timestamp, ds.timezone)"},{"reference":"src/anomalyzer/sparse_views.py","line":1,"excerpt":"\"\"\"Schema-1.1 native and complete-day evidence with independent view eligibility.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":28,"excerpt":"case.revision_id = f\"{resolution}:{case.revision_id}\""},{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/orchestrator.py","line":107,"excerpt":"if request.config.recipe == \"multi-resolution-v1\":"},{"reference":"src/anomalyzer/orchestrator.py","line":108,"excerpt":"from .sparse_views import analyze_multi_request"},{"reference":"src/anomalyzer/orchestrator.py","line":134,"excerpt":"if \"data_policy\" in source.quality and request.config.recipe != \"multi-resolution-v1\""},{"reference":"src/anomalyzer/orchestrator.py","line":170,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":195,"excerpt":"active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of)),"},{"reference":"src/anomalyzer/relationships.py","line":164,"excerpt":"source_timestamps = {source.dataset.id: source.times[source_index]"},{"reference":"src/anomalyzer/relationships.py","line":166,"excerpt":"cutoff = max(source_timestamps.values())"},{"reference":"src/anomalyzer/relationships.py","line":170,"excerpt":"source_indexes=source_indexes, source_timestamps=source_timestamps,"},{"reference":"src/anomalyzer/relationships.py","line":186,"excerpt":"lineage[-1].source_original_timestamps = {source.dataset.id: source.quality[\"samples\"][i][\"original_timestamp\"] for source, i in used}"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":124,"excerpt":"def test_sparse_phase_without_support_is_explicitly_unavailable():"},{"reference":"tests/test_sparse.py","line":181,"excerpt":"ds[\"timestamps\"][1] = \"2026-02-01T00:00:00Z\""},{"reference":"docs/sparse-data.md","line":31,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":74,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":88,"excerpt":"otherwise bounds come from the supplied as-of-available timestamp range."},{"reference":"docs/sparse-data.md","line":106,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"examples/sparse_examples.py","line":59,"excerpt":"daily = request(dataset([1.0]*60, data_policy={\"measurement\":\"additive_volume\"}), recipe=\"multi-resolution-v1\","},{"reference":"examples/sparse_examples.py","line":67,"excerpt":"\"timestamps\":[\"2026-01-01T09:00:00.123Z\"]*3+[\"2026-01-01T09:00:05.456Z\"]*4+[\"2026-01-01T09:00:10.789Z\"]*3,"},{"reference":"examples/sparse_examples.py","line":74,"excerpt":"corrected[\"datasets\"][0][\"revisions\"] = [{\"event_id\":\"call-07\", \"timestamp\":same_time[\"timestamps\"][7],"},{"reference":"examples/sparse_examples.py","line":83,"excerpt":"timestamps=same_time[\"timestamps\"][:7]+[same_time[\"timestamps\"][-1]]*12,"},{"reference":"examples/sparse_examples.py","line":96,"excerpt":"for name in (\"timestamps\", \"values\", \"event_ids\"):"},{"reference":"examples/render_readme_progression.py","line":686,"excerpt":"def figure_multi_resolution():"},{"reference":"examples/render_readme_progression.py","line":699,"excerpt":"\"config\": {\"recipe\": \"multi-resolution-v1\", \"aggregate_function\": \"sum\"},"},{"reference":"examples/render_readme_progression.py","line":710,"excerpt":"assert partial[\"included_in_completed_period_view\"] is False"},{"reference":"examples/render_readme_progression.py","line":767,"excerpt":"complete_count = result.data_quality[\"views\"][\"completed_period\"][\"observation_count\"]"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":51,"excerpt":"def valid_reset_points(cls, values):"},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":71,"excerpt":"def numeric_values(cls, data):"},{"reference":"src/anomalyzer/contracts.py","line":85,"excerpt":"if self.period_statuses is not None and len(self.period_statuses) != len(self.values):"},{"reference":"src/anomalyzer/contracts.py","line":86,"excerpt":"raise ValueError(\"period_statuses and values must have equal length\")"},{"reference":"src/anomalyzer/sparse.py","line":3,"excerpt":"Calendar coordinates remain dense only inside an explicitly bounded input grid."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":42,"excerpt":"\"\"\"Return the explicit grid, eligible model values and inspectable provenance.\"\"\""},{"reference":"src/anomalyzer/sparse.py","line":51,"excerpt":"statuses = ds.period_statuses or [\"complete\"] * len(ds.values)"},{"reference":"src/anomalyzer/sparse.py","line":57,"excerpt":"\"complete\" if value is not None else \"empty\" if policy.absence_means_zero else \"unknown\")"},{"reference":"src/anomalyzer/sparse_views.py","line":72,"excerpt":"measurements = quality[\"rule_values\"]"},{"reference":"src/anomalyzer/sparse_views.py","line":75,"excerpt":"if policy.observation_basis != \"calendar\":"},{"reference":"src/anomalyzer/sparse_views.py","line":82,"excerpt":"if policy.measurement != \"additive_volume\" or cfg.aggregate_function != \"sum\":"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":98,"excerpt":"daily_times, daily_values, coverage, records, available_at = [], [], [], [], []"},{"reference":"src/anomalyzer/orchestrator.py","line":29,"excerpt":"if not any(row.get(\"available_at\") and (row.get(\"value\") is not None or row.get(\"coverage\") == \"excluded\") for row in samples):"},{"reference":"src/anomalyzer/orchestrator.py","line":64,"excerpt":"unavailable values. Dropping only that leading run lets the ordinary engine"},{"reference":"src/anomalyzer/orchestrator.py","line":128,"excerpt":"prepared_points += len(source.values) + len(dataset.revisions)"},{"reference":"src/anomalyzer/orchestrator.py","line":140,"excerpt":"source.quality.get(\"rule_values\", source.values)[:source.quality.get(\"grid_count\", source.quality[\"observation_count\"])], dataset.rules,"},{"reference":"src/anomalyzer/orchestrator.py","line":206,"excerpt":"[dict(value=p.value, available_at=p.latest_source_cutoff if any(p.source_available_at.values()) else None) for p in lineage])"},{"reference":"src/anomalyzer/relationships.py","line":40,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/relationships.py","line":62,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":102,"excerpt":"value, unavailable = None, \"zero_denominator\""},{"reference":"src/anomalyzer/relationships.py","line":135,"excerpt":"y, x = response.values[index], predictor.values[predictor_index]"},{"reference":"src/anomalyzer/relationships.py","line":145,"excerpt":"source_values = [source.values[source_index]"},{"reference":"src/anomalyzer/early.py","line":58,"excerpt":"for index, (observed_at, value) in enumerate(zip(times, values)):"},{"reference":"src/anomalyzer/early.py","line":137,"excerpt":"criterion_met and policy.allow_explicit_rules)"},{"reference":"tests/test_sparse.py","line":45,"excerpt":"def test_zero_is_only_authorized_for_complete_empty_additive_volume():"},{"reference":"tests/test_sparse.py","line":46,"excerpt":"request = body([2, None, None, 0, 9], data_policy={\"missing\": \"zero_for_empty\", \"measurement\": \"additive_volume\"},"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":52,"excerpt":"assert rows[1][\"reason\"] == \"confirmed_empty_additive_volume\""},{"reference":"tests/test_sparse.py","line":58,"excerpt":"def test_global_collection_contract_authorizes_omitted_slots_but_not_explicit_unknown():"},{"reference":"docs/sparse-data.md","line":4,"excerpt":"Schema 1.1 therefore preserves missing values by default. An expectation can be"},{"reference":"docs/sparse-data.md","line":34,"excerpt":"data_policy=DataPolicy(),  # preserve missing; hour 2 is also unknown"},{"reference":"docs/sparse-data.md","line":42,"excerpt":"| Numeric value, including zero, `coverage=\"complete\"` | A measurement eligible for rules and model support. Explicit zero remains zero. |"},{"reference":"docs/sparse-data.md","line":44,"excerpt":"| `coverage=\"empty\"` | Caller confirms a finalized period with no reported volume. Preserve as missing unless an additive zero policy explicitly authorizes conversion. |"},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"unknown\"` | Collection/exposure is unknown; never filled even under a global empty-volume contract. |"},{"reference":"examples/sparse_examples.py","line":22,"excerpt":"preserve = request(dataset(values, coverage=[\"empty\" if v is None else \"complete\" for v in values],"},{"reference":"examples/sparse_examples.py","line":23,"excerpt":"data_policy={\"measurement\": \"additive_volume\"}))"},{"reference":"examples/sparse_examples.py","line":25,"excerpt":"zero[\"datasets\"][0][\"data_policy\"][\"missing\"] = \"zero_for_empty\""},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":28,"excerpt":"coverage=[\"complete\"]*2+[\"unknown\"]+[\"complete\"]*6+[\"unknown\"]+[\"complete\"]*3+[\"incomplete\"], data_policy={}))"},{"reference":"examples/render_readme_progression.py","line":135,"excerpt":"def chart(svg, values, overlays, x, y, width, height, *, y_bounds=None,"},{"reference":"examples/render_readme_progression.py","line":139,"excerpt":"all_values = [values] + [series for _, series, *_ in overlays]"},{"reference":"examples/render_readme_progression.py","line":140,"excerpt":"low, high = y_bounds or bounds(*all_values)"},{"reference":"examples/render_readme_progression.py","line":166,"excerpt":"if value is not None and (index == 0 or values[index-1] is None) and (index == count-1 or values[index+1] is None):"},{"reference":"examples/render_readme_progression.py","line":206,"excerpt":"values = [80 + 0.65 * index + [0, 0.25, -0.15, 0.10, -0.20][index % 5]"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":242,"excerpt":"missing: Literal[\"preserve\", \"zero_for_empty\"] = \"preserve\""},{"reference":"src/anomalyzer/contracts.py","line":245,"excerpt":"observation_basis: Literal[\"calendar\", \"event_index\"] = \"calendar\""},{"reference":"src/anomalyzer/contracts.py","line":246,"excerpt":"event_order: Literal[\"timestamp\", \"trusted_sequence\"] = \"timestamp\""},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":52,"excerpt":"for index, (text, value, status) in enumerate(zip(ds.timestamps, ds.values, statuses)):"},{"reference":"src/anomalyzer/sparse.py","line":60,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":66,"excerpt":"if cfg.duplicate_policy == \"error\" or policy.observation_basis == \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":67,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":128,"excerpt":"policy_zero_children=sum(quality[\"samples\"][i][\"reason\"] == \"confirmed_empty_additive_volume\" for i in indexes),"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":134,"excerpt":"timestamps=daily_times, values=daily_values, frequency=\"1d\", coverage=coverage, available_at=available_at,"},{"reference":"src/anomalyzer/orchestrator.py","line":38,"excerpt":"assessment.baseline[\"known_input_availability_cutoff\"] = cutoffs[assessment.index]"},{"reference":"src/anomalyzer/orchestrator.py","line":39,"excerpt":"assessment.assumptions.append(\"Availability cutoff is conservative across known prior inputs; unknown arrival history cannot establish delivery-time detection performance.\")"},{"reference":"src/anomalyzer/orchestrator.py","line":170,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":191,"excerpt":"relationship.id, derived.timestamps or [], derived.values,"},{"reference":"src/anomalyzer/orchestrator.py","line":195,"excerpt":"active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of)),"},{"reference":"src/anomalyzer/relationships.py","line":44,"excerpt":"raise ValueError(\"relationship timestamps cannot be positional\")"},{"reference":"src/anomalyzer/relationships.py","line":61,"excerpt":"for index in range(relationship.lag, coefficient_cutoff + 1):"},{"reference":"src/anomalyzer/relationships.py","line":62,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":100,"excerpt":"if relationship.zero_denominator == \"error\":"},{"reference":"src/anomalyzer/relationships.py","line":101,"excerpt":"raise ValueError(f\"zero denominator at aligned index {index}\")"},{"reference":"src/anomalyzer/early.py","line":140,"excerpt":"unavailable_rules = [dict(id=rule.id, reason=\"no_comparable_prior\" if previous is None else \"relative_change_from_zero\")"},{"reference":"src/anomalyzer/early.py","line":147,"excerpt":"target_id=target_id, index=index, timestamp=observed_at,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":68,"excerpt":"{\"observation_basis\": \"event_index\", \"grid_start\": \"2026-01-01T00:00:00Z\"},"},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":129,"excerpt":"assert any(row[\"reason\"] == \"no_training_support_for_calendar_phase\" for row in result.methods[0].diagnostics[\"unavailable\"])"},{"reference":"docs/sparse-data.md","line":31,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":106,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"docs/sparse-data.md","line":111,"excerpt":"rejected. Event-index timestamps without `event_ids` must be unique; identified"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":125,"excerpt":"timestamps=[\"2026-01-01T09:00:10.789Z\", \"2026-01-01T09:00:10.789Z\"],"},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":38,"excerpt":"events[\"timestamps\"] = [(datetime(2026,1,1,tzinfo=timezone.utc)+timedelta(hours=h)).isoformat() for h in hours]"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":67,"excerpt":"\"timestamps\":[\"2026-01-01T09:00:00.123Z\"]*3+[\"2026-01-01T09:00:05.456Z\"]*4+[\"2026-01-01T09:00:10.789Z\"]*3,"},{"reference":"examples/render_readme_progression.py","line":654,"excerpt":"svg.text(430, 467, \"Sample index (zero-based)\", 15, MUTED, anchor=\"middle\")"},{"reference":"examples/render_readme_progression.py","line":697,"excerpt":"\"timestamps\": [(origin + timedelta(hours=index)).isoformat()"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"},{"reference":"examples/render_readme_progression.py","line":887,"excerpt":"shared_high = max(v for v in observed(\"preserve\")+expectation(\"preserve\")+observed(\"zero_empty\")+expectation(\"zero_empty\") if v is not None)*1.12"},{"reference":"examples/render_readme_progression.py","line":897,"excerpt":"header(svg, \"15\", \"Keep zero, unknown and unfinished distinct\", \"Blue zeros are actual measurements; gaps stay gaps; the orange open-period value is excluded.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","order":30}},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","status":"implemented","area":"data","path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","health":"review","coverage":"partial","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","order":10}},{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","status":"implemented","area":"data","path":".capabilities/data/quality/identify-distinct-events.capability.yaml","health":"review","coverage":"partial","summary":"Callers supply stable logical event IDs so distinct comparable calls at one timestamp retain separate values and evidence without inventing chronology.","intent":"Make real per-call evidence consumable while protecting references, replay and pattern interpretation from retry duplication and arbitrary tie order.","references":{"total":12,"readable":12,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":308,"excerpt":"identified = self.event_ids is not None"},{"reference":"src/anomalyzer/contracts.py","line":315,"excerpt":"raise ValueError(\"identified event revisions require event_id\")"},{"reference":"src/anomalyzer/contracts.py","line":319,"excerpt":"if trusted and (not identified or self.event_sequence is None):"},{"reference":"src/anomalyzer/contracts.py","line":596,"excerpt":"dataset_results: list[DatasetResultV11]"},{"reference":"src/anomalyzer/contracts.py","line":597,"excerpt":"relationship_results: list[RelationshipResult]"},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":14,"excerpt":"CONTRACT = \"event-identity-v1\""},{"reference":"src/anomalyzer/events.py","line":100,"excerpt":"key, identity = (revision.event_id, available), (revision.event_id, revision.revision_id)"},{"reference":"src/anomalyzer/events.py","line":102,"excerpt":"raise ValueError(\"duplicate event revision identity or availability\")"},{"reference":"src/anomalyzer/events.py","line":165,"excerpt":"arrival_semantics=\"supplied availability, otherwise event-time snapshot; no delivery latency claim\","},{"reference":"src/anomalyzer/sparse.py","line":182,"excerpt":"arrival_semantics=\"supplied availability timestamps; unannotated samples are event-time snapshots\" if any(s[\"available_at\"] and (s[\"value\"] is not None or s[\"coverage\"] == \"excluded\") for s in samples) else"},{"reference":"src/anomalyzer/sparse.py","line":194,"excerpt":"identified = \"event_identity_contract\" in quality"},{"reference":"src/anomalyzer/sparse.py","line":195,"excerpt":"resets = quality[\"event_reset_positions\"] if identified else active_resets(cfg.reset_points, times, ds.timezone, context.as_of)"},{"reference":"src/anomalyzer/sparse.py","line":202,"excerpt":"\"profile_revision\": \"identified-events-v1\" if identified else \"sparse-frozen-phase-v1\","},{"reference":"src/anomalyzer/sparse.py","line":333,"excerpt":"[method.model_copy(update={\"evidence\": [e]})], ds.id)) if identified else episodes([temporary], ds.id)"},{"reference":"src/anomalyzer/orchestrator.py","line":122,"excerpt":"dataset_results = []"},{"reference":"src/anomalyzer/orchestrator.py","line":143,"excerpt":"if \"event_identity_contract\" in source.quality else"},{"reference":"src/anomalyzer/orchestrator.py","line":155,"excerpt":"dataset_results.append(DatasetResultV11("},{"reference":"src/anomalyzer/orchestrator.py","line":162,"excerpt":"dataset_results.append(_empty_dataset_result(dataset.id, exc))"},{"reference":"src/anomalyzer/orchestrator.py","line":164,"excerpt":"relationship_results = []"},{"reference":"src/anomalyzer/alignment.py","line":63,"excerpt":"if any(\"event_identity_contract\" in source.quality for source in sources):"},{"reference":"src/anomalyzer/alignment.py","line":66,"excerpt":"if (not all(\"event_identity_contract\" in source.quality for source in sources)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/relationships.py","line":198,"excerpt":"event_order=policies[0].event_order if identified else \"timestamp\","},{"reference":"src/anomalyzer/relationships.py","line":203,"excerpt":"event_ids=[s[\"event_id\"] for s in sources[0].quality[\"samples\"]] if identified else None,"},{"reference":"src/anomalyzer/relationships.py","line":205,"excerpt":"event_sequence=[s[\"event_sequence\"] for s in sources[0].quality[\"samples\"]] if identified and policies[0].event_order == \"trusted_sequence\" else None)"},{"reference":"src/anomalyzer/cases.py","line":29,"excerpt":"def compose_cases(dataset_results: list[DatasetResultV11],"},{"reference":"src/anomalyzer/cases.py","line":30,"excerpt":"relationship_results: list[RelationshipResult],"},{"reference":"src/anomalyzer/cases.py","line":33,"excerpt":"for result in dataset_results:"},{"reference":"src/anomalyzer/cases.py","line":37,"excerpt":"for result in relationship_results:"},{"reference":"src/anomalyzer/cases.py","line":62,"excerpt":"case_id = \"case:\" + (case_digest if event_id is not None else case_digest[:16])"},{"reference":"src/anomalyzer/mcp_views.py","line":12,"excerpt":"for item in result[\"dataset_results\"]:"},{"reference":"src/anomalyzer/mcp_views.py","line":14,"excerpt":"for item in result[\"relationship_results\"]:"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":77,"excerpt":"if \"event_identity_contract\" in source else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":88,"excerpt":"def _case_assessments(result, case):"},{"reference":"tests/test_events.py","line":1,"excerpt":"\"\"\"Logical event identity, honest tie semantics and deterministic consumer replay.\"\"\""},{"reference":"tests/test_events.py","line":12,"excerpt":"from anomalyzer import analyze, analyze_series, get_case, replay_policy"},{"reference":"tests/test_events.py","line":36,"excerpt":"return [item for method in result.dataset_results[0].methods for item in method.evidence]"},{"reference":"tests/test_events.py","line":49,"excerpt":"item = result.dataset_results[0]"},{"reference":"tests/test_events.py","line":51,"excerpt":"assert item.data_quality[\"event_identity_contract\"] == \"event-identity-v1\""},{"reference":"docs/sparse-data.md","line":37,"excerpt":"print(result.dataset_results[0].data_quality[\"samples\"])"},{"reference":"docs/sparse-data.md","line":97,"excerpt":"Full evidence can be several megabytes for a bounded two-source history; the"},{"reference":"docs/sparse-data.md","line":100,"excerpt":"and result bytes for 1,344-hour and bounded event workloads. CI runs it in the"},{"reference":"docs/sparse-data.md","line":111,"excerpt":"rejected. Event-index timestamps without `event_ids` must be unique; identified"},{"reference":"docs/sparse-data.md","line":115,"excerpt":"## Stable event identities since 0.3.0"},{"reference":"examples/sparse_examples.py","line":39,"excerpt":"result[\"cost_events\"] = request(events)"},{"reference":"examples/sparse_examples.py","line":72,"excerpt":"result[\"identified_calls\"] = request(same_time, training_size=2, calibration_size=3)"},{"reference":"examples/sparse_examples.py","line":73,"excerpt":"corrected = deepcopy(result[\"identified_calls\"])"},{"reference":"examples/sparse_examples.py","line":77,"excerpt":"result[\"identified_corrected\"] = corrected"},{"reference":"examples/sparse_examples.py","line":80,"excerpt":"result[\"identified_before_correction\"] = before"},{"reference":"examples/render_readme_progression.py","line":3,"excerpt":"The figures use checked-in synthetic examples and real ``analyze`` results."},{"reference":"examples/render_readme_progression.py","line":781,"excerpt":"source_results = {item.dataset_id: item for item in result.dataset_results}"},{"reference":"examples/render_readme_progression.py","line":782,"excerpt":"relationship = result.relationship_results[0]"},{"reference":"examples/render_readme_progression.py","line":787,"excerpt":"assert not any(row[\"triggers\"] for item in result.dataset_results"},{"reference":"examples/render_readme_progression.py","line":798,"excerpt":"for row in source_results[dataset_id].methods[0].evidence}"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":246,"excerpt":"event_order: Literal[\"timestamp\", \"trusted_sequence\"] = \"timestamp\""},{"reference":"src/anomalyzer/contracts.py","line":255,"excerpt":"if self.event_order == \"trusted_sequence\" and self.observation_basis != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":256,"excerpt":"raise ValueError(\"trusted_sequence requires event_index\")"},{"reference":"src/anomalyzer/contracts.py","line":268,"excerpt":"EventId = Annotated[str, Field(min_length=1, max_length=128, strict=True, pattern=r\"^\\S(?:.*\\S)?$\")]"},{"reference":"src/anomalyzer/contracts.py","line":296,"excerpt":"event_sequence: list[Annotated[int, Field(ge=0, le=9223372036854775807, strict=True)]] | None = Field(default=None, json_schema_extra={\"uniqueItems\": True})"},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":50,"excerpt":"if policy.event_order != \"trusted_sequence\" and times[index] == times[index-1]:"},{"reference":"src/anomalyzer/events.py","line":71,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/events.py","line":80,"excerpt":"event_sequence=ds.event_sequence[index] if ds.event_sequence is not None else None,"},{"reference":"src/anomalyzer/events.py","line":84,"excerpt":"if policy.event_order == \"trusted_sequence\":"},{"reference":"src/anomalyzer/sparse.py","line":46,"excerpt":"if ds.event_ids is not None:"},{"reference":"src/anomalyzer/sparse.py","line":62,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/sparse.py","line":93,"excerpt":"raise ValueError(\"revision available_at cannot precede event time\")"},{"reference":"src/anomalyzer/sparse.py","line":170,"excerpt":"regularity_basis=\"declared_cadence\" if policy.observation_basis == \"calendar\" else \"explicit_comparable_events\","},{"reference":"src/anomalyzer/sparse.py","line":182,"excerpt":"arrival_semantics=\"supplied availability timestamps; unannotated samples are event-time snapshots\" if any(s[\"available_at\"] and (s[\"value\"] is not None or s[\"coverage\"] == \"excluded\") for s in samples) else"},{"reference":"src/anomalyzer/orchestrator.py","line":24,"excerpt":"from .events import annotate_assessments"},{"reference":"src/anomalyzer/orchestrator.py","line":142,"excerpt":"reset_positions=(source.quality[\"event_reset_positions\"] + source.quality[\"sequence_boundaries\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":193,"excerpt":"reset_positions=(result.data_quality[\"event_reset_positions\"] + result.data_quality[\"sequence_boundaries\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":209,"excerpt":"if assessment.event_id is not None:"},{"reference":"src/anomalyzer/orchestrator.py","line":213,"excerpt":"\"samples\", \"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")})"},{"reference":"src/anomalyzer/alignment.py","line":64,"excerpt":"signatures = [[(s.get(\"event_id\"), s.get(\"event_sequence\"))"},{"reference":"src/anomalyzer/alignment.py","line":66,"excerpt":"if (not all(\"event_identity_contract\" in source.quality for source in sources)"},{"reference":"src/anomalyzer/alignment.py","line":68,"excerpt":"or len({source.quality[\"event_order\"] for source in sources}) != 1):"},{"reference":"src/anomalyzer/relationships.py","line":38,"excerpt":"raise ValueError(\"lagged_response cannot order tied events without trusted_sequence\")"},{"reference":"src/anomalyzer/relationships.py","line":198,"excerpt":"event_order=policies[0].event_order if identified else \"timestamp\","},{"reference":"src/anomalyzer/relationships.py","line":205,"excerpt":"event_sequence=[s[\"event_sequence\"] for s in sources[0].quality[\"samples\"]] if identified and policies[0].event_order == \"trusted_sequence\" else None)"},{"reference":"src/anomalyzer/cases.py","line":59,"excerpt":"\"relationships\": relationships, **({\"event_id\": event_id} if event_id is not None else {})},"},{"reference":"src/anomalyzer/cases.py","line":62,"excerpt":"case_id = \"case:\" + (case_digest if event_id is not None else case_digest[:16])"},{"reference":"src/anomalyzer/cases.py","line":115,"excerpt":"for _, _, assessment in sorted(items, key=lambda entry: entry[2].id)]} if event_id is not None else {}),"},{"reference":"src/anomalyzer/cases.py","line":118,"excerpt":"revision_id = case_id + \":\" + (revision_digest if event_id is not None else revision_digest[:12])"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":152,"excerpt":"**({\"event_id\": case[\"event_id\"]} if case.get(\"event_id\") is not None else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":380,"excerpt":"if point.get(\"event_id\") is not None and point[\"index\"] in ("},{"reference":"src/anomalyzer/mcp_views.py","line":385,"excerpt":"\"note\": \"Evidence establishes numerical criteria only; source context, cause, impact, and action authority require external review.\","},{"reference":"tests/test_events.py","line":29,"excerpt":"ds[\"event_sequence\"] = list(range(100, 100+len(values)))"},{"reference":"tests/test_events.py","line":30,"excerpt":"ds[\"data_policy\"][\"event_order\"] = \"trusted_sequence\""},{"reference":"tests/test_events.py","line":80,"excerpt":"lambda d: d.update(event_sequence=list(range(7))),"},{"reference":"tests/test_events.py","line":81,"excerpt":"lambda d: d[\"data_policy\"].update(event_order=\"trusted_sequence\"),"},{"reference":"tests/test_events.py","line":121,"excerpt":"def test_trusted_sequence_is_explicit_and_enables_order_without_changing_time():"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":136,"excerpt":"| `event_ids` | Optional array aligned with original values/timestamps; every ID is a unique, case-sensitive string of 1–128 characters with no surrounding whitespace or line breaks. IDs require explicit event-index policy, original timestamps and no frequency. |"},{"reference":"docs/sparse-data.md","line":137,"excerpt":"| `data_policy.event_order` | `\"timestamp\"` by default. Sort by normalized time then exact ID for deterministic layout; equal-time ID order is **not** temporal order. |"},{"reference":"docs/sparse-data.md","line":138,"excerpt":"| `event_sequence` | Optional aligned array of unique strict integers in 0…9223372036854775807. Requires `event_order=\"trusted_sequence\"` and IDs. Ordinals must agree with nondecreasing original timestamps. Gaps in ordinals are allowed and do not represent missing events or elapsed time. |"},{"reference":"docs/sparse-data.md","line":139,"excerpt":"| `SampleRevision.event_id` | Required for every revision of an identified source. It addresses an existing ID at exactly the same original timestamp. A revision changes value/coverage, never event identity, time or sequence. |"},{"reference":"examples/sparse_examples.py","line":87,"excerpt":"ordered[\"event_sequence\"] = list(range(100,119))"},{"reference":"examples/sparse_examples.py","line":88,"excerpt":"ordered[\"data_policy\"][\"event_order\"] = \"trusted_sequence\""},{"reference":"examples/sparse_examples.py","line":89,"excerpt":"result[\"identified_pattern_sequence\"] = request(ordered, training_size=2, calibration_size=3, point_threshold=100)"},{"reference":"examples/render_readme_progression.py","line":918,"excerpt":"header(svg, \"17\", \"Compare real calls across quiet hours\", \"A per-call value is undefined in a no-call hour. Event position is an explicitly different observation basis.\")"},{"reference":"examples/render_readme_progression.py","line":921,"excerpt":"svg.text(95, 585, \"Event 0: Jan 1 00:00 UTC  ·  event 11: Jan 18 22:00 UTC  ·  no elapsed-time rate inferred\", 16, MUTED)"},{"reference":"examples/render_readme_progression.py","line":1024,"excerpt":"(\"identified_pattern_sequence\",410,\"Trusted sequence: ordered residual checks become applicable\")):"},{"reference":"examples/render_readme_progression.py","line":1037,"excerpt":"svg.text(60,665,\"Only assert trusted_sequence from actual source order. Neither sequence nor distinct IDs proves independence or cause.\",15,MUTED)"},{"reference":"examples/render_readme_progression.py","line":1038,"excerpt":"figures[\"anomaly-progression-24-event-order\"] = svg.finish()"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":118,"excerpt":"score_semantics: str"},{"reference":"src/anomalyzer/contracts.py","line":192,"excerpt":"# This avoids changing the generated 1.0 schemas while giving agent-facing"},{"reference":"src/anomalyzer/contracts.py","line":193,"excerpt":"# callers explicit evidence and relationship semantics."},{"reference":"src/anomalyzer/contracts.py","line":241,"excerpt":"\"\"\"Explicit source semantics; unknown collection is never implicit zero.\"\"\""},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":14,"excerpt":"CONTRACT = \"event-identity-v1\""},{"reference":"src/anomalyzer/events.py","line":22,"excerpt":"def event_token(event_id, entity):"},{"reference":"src/anomalyzer/events.py","line":23,"excerpt":"return fingerprint(dict(event_id=event_id, entity=entity))"},{"reference":"src/anomalyzer/events.py","line":26,"excerpt":"def event_resets(points, times, policy, zone, as_of):"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":23,"excerpt":"from .validation import resolve_reset_points, timestamp"},{"reference":"src/anomalyzer/sparse.py","line":46,"excerpt":"if ds.event_ids is not None:"},{"reference":"src/anomalyzer/sparse.py","line":60,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":62,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/orchestrator.py","line":21,"excerpt":"from .relationships import derive, _source_ids"},{"reference":"src/anomalyzer/orchestrator.py","line":84,"excerpt":"evidence_ids = {}"},{"reference":"src/anomalyzer/orchestrator.py","line":90,"excerpt":"evidence_ids[old] = item[\"id\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":94,"excerpt":"observation[\"evidence_refs\"] = [evidence_ids.get(value, value)"},{"reference":"src/anomalyzer/orchestrator.py","line":100,"excerpt":"pattern[\"evidence_refs\"] = [evidence_ids.get(value, value)"},{"reference":"src/anomalyzer/alignment.py","line":14,"excerpt":"source_indexes: list[int]"},{"reference":"src/anomalyzer/alignment.py","line":33,"excerpt":"indexes = ([sample[\"source_index\"] if len(sample[\"source_indexes\"]) <= 1 else -1"},{"reference":"src/anomalyzer/alignment.py","line":46,"excerpt":"def exact_sources(ids: list[str], prepared: dict[str, PreparedDataset]):"},{"reference":"src/anomalyzer/alignment.py","line":47,"excerpt":"sources = [prepared[source_id] for source_id in ids]"},{"reference":"src/anomalyzer/alignment.py","line":48,"excerpt":"if any(source.dataset.timestamps is None for source in sources):"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/cases.py","line":16,"excerpt":"source_id = relationship.numerator"},{"reference":"src/anomalyzer/cases.py","line":18,"excerpt":"source_id = relationship.minuend"},{"reference":"src/anomalyzer/cases.py","line":20,"excerpt":"source_id = relationship.observed"},{"reference":"src/anomalyzer/cases.py","line":22,"excerpt":"source_id = relationship.response"},{"reference":"src/anomalyzer/cases.py","line":24,"excerpt":"source_id = relationship.conditions[0].dataset"},{"reference":"src/anomalyzer/mcp_views.py","line":68,"excerpt":"\"latest_evidence_state\": source.get(\"latest_evidence_state\"),"},{"reference":"src/anomalyzer/mcp_views.py","line":71,"excerpt":"\"grid_count\": source[\"grid_count\"], \"usable_model_count\": source[\"usable_model_count\"],"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":77,"excerpt":"if \"event_identity_contract\" in source else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":89,"excerpt":"ids = set(case[\"supporting_evidence\"])"},{"reference":"tests/test_events.py","line":1,"excerpt":"\"\"\"Logical event identity, honest tie semantics and deterministic consumer replay.\"\"\""},{"reference":"tests/test_events.py","line":22,"excerpt":"ds = dict(id=\"cost\", entity={\"application\": \"synthetic\"}, units=\"dollars_per_call\","},{"reference":"tests/test_events.py","line":24,"excerpt":"event_ids=[f\"call-{i:04}\" for i in range(len(values))],"},{"reference":"tests/test_events.py","line":25,"excerpt":"data_policy={\"observation_basis\": \"event_index\"},"},{"reference":"tests/test_events.py","line":41,"excerpt":"return {k: semantic(v) for k, v in value.items() if k not in (\"run_id\", \"runtime_seconds\", \"source_index\", \"source_indexes\")}"},{"reference":"docs/sparse-data.md","line":50,"excerpt":"`coverage`, `available_at` and legacy `period_statuses` align with the original"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":74,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":77,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":83,"excerpt":"## Calendar coordinates and bounded grids"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":65,"excerpt":"# Genuine distinct calls share the recorded timestamp; IDs do not order them."},{"reference":"examples/sparse_examples.py","line":66,"excerpt":"same_time = {\"id\":\"call-cost\", \"entity\":{\"application\":\"synthetic\"}, \"units\":\"dollars_per_call\","},{"reference":"examples/sparse_examples.py","line":69,"excerpt":"\"event_ids\":[f\"call-{i:02}\" for i in range(10)],"},{"reference":"examples/render_readme_progression.py","line":141,"excerpt":"sx = lambda index: x + index / max(count - 1, 1) * width"},{"reference":"examples/render_readme_progression.py","line":301,"excerpt":"frozen_flags = {row[\"index\"] for row in frozen.methods[0].evidence"},{"reference":"examples/render_readme_progression.py","line":303,"excerpt":"adaptive_flags = {row[\"index\"] for row in adaptive.methods[0].evidence"},{"reference":"examples/render_readme_progression.py","line":373,"excerpt":"sx = lambda index: x + index / (len(sample) - 1) * width"},{"reference":"examples/render_readme_progression.py","line":413,"excerpt":"scores = {row[\"index\"]: row[\"standardized_residual\"] for row in evidence}"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","status":"implemented","area":"data","path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","health":"review","coverage":"partial","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","references":{"total":9,"readable":9,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar duplicate timestamps require an aggregation policy; identified event-index calls can share original timestamps while duplicate logical IDs are rejected and arbitrary tie order supplies no sequence evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","order":20}},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","status":"implemented","area":"execution","path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","health":"review","coverage":"partial","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","order":50}},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","health":"review","coverage":"partial","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","order":50}},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","health":"review","coverage":"partial","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Readiness status remains visible, including when it was not assessed.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","order":60}},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","health":"review","coverage":"partial","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Analysis tool calls return structured content together with compact text.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","order":10}},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","order":20}},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","health":"review","coverage":"partial","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","order":10}},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","health":"review","coverage":"partial","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","references":{"total":11,"readable":11,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/publish_release.py","line":54,"excerpt":"\"Linux/Python 3.12 Lambda consumer verified offline. \""},{"reference":"scripts/verify_distribution.py","line":47,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":6,"excerpt":"bundle=\"$root/.tmp/lambda-consumer\""},{"reference":"scripts/verify_lambda.sh","line":18,"excerpt":"python -m pip download --only-binary=:all: --require-hashes \\"},{"reference":"scripts/verify_lambda.sh","line":20,"excerpt":"python -m pip install --no-index --only-binary=:all: --require-hashes \\"},{"reference":".github/workflows/ci.yml","line":29,"excerpt":"- run: uv run python examples/generate.py"},{"reference":".github/workflows/ci.yml","line":32,"excerpt":"- run: uv run python examples/run_demo.py"},{"reference":".github/workflows/ci.yml","line":33,"excerpt":"- run: uv run python examples/run_agentic_demo.py"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/ci.yml","line":54,"excerpt":"run: uv export --locked --no-dev --no-emit-project --output-file dist/runtime-dependencies.txt"},{"reference":".github/workflows/release.yml","line":62,"excerpt":"- name: Verify the downloaded release in the offline Lambda-compatible bundle"},{"reference":"docs/releases/README.md","line":16,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":62,"excerpt":"uv run python examples/run_demo.py"},{"reference":"docs/releases/README.md","line":63,"excerpt":"uv run python examples/run_agentic_demo.py"},{"reference":"docs/releases/README.md","line":79,"excerpt":"versions, and tests a bundled Linux x86_64/Python 3.12 consumer in AWS's Lambda"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","evidence":[{"reference":"scripts/verify_distribution.py","line":1,"excerpt":"\"\"\"Install wheel AND sdist in clean environments outside the source tree (requires uv).\"\"\""},{"reference":"scripts/verify_distribution.py","line":48,"excerpt":"run([\"uv\", \"pip\", \"install\", \"--python\", python, \"--no-deps\", artifact], outside, env=env)"},{"reference":"examples/library/guard_smoke.py","line":32,"excerpt":"analyze_relationships, get_case, replay_policy,"},{"reference":"examples/library/guard_smoke.py","line":49,"excerpt":"assert legacy.methods[0].diagnostics[\"detection_readiness\"]"},{"reference":"examples/library/guard_smoke.py","line":65,"excerpt":"\"relationships\": [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"cost\","},{"reference":"examples/library/guard_smoke.py","line":69,"excerpt":"result = analyze_relationships(request)"},{"reference":"examples/library/guard_smoke.py","line":76,"excerpt":"short = analyze_relationships({**request.model_dump(mode=\"json\"),"},{"reference":".github/workflows/ci.yml","line":59,"excerpt":"- name: Test built wheel and sdist outside checkout"},{"reference":".github/workflows/release.yml","line":60,"excerpt":"- name: Install the exact published wheel URL with hashes outside checkout"},{"reference":"docs/releases/README.md","line":17,"excerpt":"| Source distribution (`.tar.gz`, or sdist) | Source needed to build the package. Useful for rebuilding and inspection; installation also needs the declared build backend. |"},{"reference":"docs/releases/README.md","line":99,"excerpt":"assets, installs the exact public wheel URL with its hashes in a clean consumer,"},{"reference":"docs/releases/README.md","line":217,"excerpt":"`analyze_series`, `analyze_relationships`, `get_case`, `replay_policy` (also"},{"reference":"docs/releases/README.md","line":220,"excerpt":"`anomalyzer.contracts`. See the self-contained"},{"reference":"docs/releases/README.md","line":222,"excerpt":"all APIs, recipes, point/pattern evidence, relationships, cases and policy"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":34,"excerpt":"return hashlib.sha256(path.read_bytes()).hexdigest()"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/release.py","line":85,"excerpt":"suffix = f\" --hash=sha256:{digest(wheel)}\\n\""},{"reference":"scripts/release.py","line":92,"excerpt":"assets = sorted(p for p in dist.iterdir() if p.is_file()"},{"reference":"scripts/publish_release.py","line":1,"excerpt":"\"\"\"Publish tested assets without replacing tags or existing assets (GitHub CLI auth).\"\"\""},{"reference":"scripts/publish_release.py","line":18,"excerpt":"def find_release(tag):"},{"reference":"scripts/publish_release.py","line":19,"excerpt":"\"\"\"The by-tag endpoint omits drafts; the authenticated release list includes them.\"\"\""},{"reference":"scripts/publish_release.py","line":20,"excerpt":"endpoint = f\"repos/{REPOSITORY}/releases/tags/{tag}\""},{"reference":"scripts/publish_release.py","line":40,"excerpt":"remote_sha = json.loads(gh(\"api\", f\"repos/{REPOSITORY}/commits/{tag}\"))[\"sha\"]"},{"reference":"scripts/verify_distribution.py","line":47,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":25,"excerpt":"sha256sum *.whl > /output/wheelhouse-SHA256SUMS"},{"reference":"examples/library/guard_smoke.py","line":39,"excerpt":"assert anomalyzer.__version__ == importlib.metadata.version(\"anomalyzer\")"},{"reference":".github/workflows/ci.yml","line":44,"excerpt":"- name: Check release tag before building"},{"reference":".github/workflows/ci.yml","line":47,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/ci.yml","line":48,"excerpt":"run: python scripts/release.py check-version --tag \"$RELEASE_TAG\""},{"reference":".github/workflows/ci.yml","line":51,"excerpt":"- name: Check package metadata"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/release.yml","line":34,"excerpt":"name: release-dist-${{ github.sha }}"},{"reference":".github/workflows/release.yml","line":36,"excerpt":"- name: Stage tested assets for owner publication"},{"reference":".github/workflows/release.yml","line":39,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/release.yml","line":40,"excerpt":"run: python scripts/publish_release.py --tag \"$RELEASE_TAG\" --revision \"$GITHUB_SHA\""},{"reference":".github/workflows/release.yml","line":48,"excerpt":"ref: ${{ github.event.release.tag_name }}"},{"reference":"docs/releases/README.md","line":8,"excerpt":"The original `v0.2.0` tag remains an unpublished candidate after a draft-lookup"},{"reference":"docs/releases/README.md","line":16,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":18,"excerpt":"| Version tag (`v0.3.0`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":23,"excerpt":"it for package metadata and the public `__version__` imports it."},{"reference":"tests/test_release.py","line":1,"excerpt":"\"\"\"Release safety: reject wrong tags and refuse changes to published artifacts.\"\"\""},{"reference":"tests/test_release.py","line":21,"excerpt":"def test_release_refuses_mismatched_or_unsupported_tag(tag):"},{"reference":"tests/test_release.py","line":23,"excerpt":"release.check_tag(tag)"},{"reference":"tests/test_release.py","line":52,"excerpt":"def test_published_matching_release_is_noop(monkeypatch, tmp_path):"},{"reference":"tests/test_release.py","line":81,"excerpt":"\"version\": release.version(), \"source_revision\": \"a\" * 40, \"tag\": f\"v{release.version()}\","}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","order":20}},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","health":"review","coverage":"partial","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","order":10}},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","health":"review","coverage":"partial","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","order":20}},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","status":"implemented","area":"investigation","path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","health":"review","coverage":"partial","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","order":10}},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","status":"implemented","area":"investigation","path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","order":40}},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","status":"implemented","area":"investigation","path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","health":"review","coverage":"partial","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Default policy leaves action and notification eligibility disabled.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","order":60}},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","status":"implemented","area":"investigation","path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","health":"review","coverage":"partial","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","order":30}},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","health":"review","coverage":"partial","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","order":20}},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","health":"review","coverage":"partial","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","order":30}},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","status":"implemented","area":"relationships","path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","health":"review","coverage":"partial","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Difference relationships require identical declared source units and matching output units.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","order":10}},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","status":"implemented","area":"relationships","path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","health":"review","coverage":"partial","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":22,"excerpt":"cases = compose_cases([d for d in datasets if d.resolution == resolution],"},{"reference":"src/anomalyzer/sparse_views.py","line":146,"excerpt":"native_sources = [d.model_copy(update={\"data_policy\": d.data_policy or DataPolicy()}) for d in request.datasets]"},{"reference":"src/anomalyzer/sparse_views.py","line":149,"excerpt":"native_request = request.model_copy(update={\"datasets\": native_sources, \"config\": native_config})"},{"reference":"src/anomalyzer/sparse_views.py","line":169,"excerpt":"daily_request = request.model_copy(update={\"datasets\": sources, \"config\": daily_config, \"relationships\": daily_relationships})"},{"reference":"src/anomalyzer/agent.py","line":14,"excerpt":"\"\"\"Use legacy semantics for a Dataset; explicit V11 fields select source policies.\"\"\""},{"reference":"src/anomalyzer/agent.py","line":20,"excerpt":"name in dataset for name in (\"data_policy\", \"coverage\", \"available_at\", \"revisions\", \"rules\", \"completed_period_rules\", \"event_ids\", \"event_sequence\")):"},{"reference":"src/anomalyzer/agent.py","line":23,"excerpt":"return analyze({\"schema_version\": \"1.1\", \"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"src/anomalyzer/agent.py","line":29,"excerpt":"return analyze({\"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"tests/test_sparse.py","line":18,"excerpt":"return {\"schema_version\": \"1.1\", \"datasets\": [{\"id\": \"series\", \"entity\": {\"customer\": \"test\"},"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":88,"excerpt":"def test_declared_duplicate_aggregation_preserves_provenance(policy, value):"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":118,"excerpt":"changed[\"datasets\"][0][\"values\"][-10:] = [999]*10"},{"reference":"docs/sparse-data.md","line":36,"excerpt":"result = analyze({\"schema_version\": \"1.1\", \"datasets\": [source.model_dump(mode=\"json\")]})"},{"reference":"docs/sparse-data.md","line":117,"excerpt":"Schema 1.1 adds the optional **`event-identity-v1`** contract. It is selected by"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":156,"excerpt":"`data_quality.event_identity_contract=\"event-identity-v1\"` record those semantics."},{"reference":"docs/sparse-data.md","line":206,"excerpt":"and a selected-input fingerprint. Evidence IDs use stable event identity within"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":70,"excerpt":"\"\"\"Sum only wholly observed finalized additive children. Never average ratios.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":175,"excerpt":"applicability=\"completed-day view supports ratio-of-sums and differences of compatible additive sums only\","},{"reference":"src/anomalyzer/relationships.py","line":94,"excerpt":"used = [(numerator, index), (denominator, index)]"},{"reference":"src/anomalyzer/relationships.py","line":177,"excerpt":"\"complete\" if source.values[i] is not None else \"unknown_or_incomplete\" for source, i in used}"},{"reference":"tests/test_sparse.py","line":231,"excerpt":"request[\"relationships\"] = [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"series\", \"denominator\": \"calls\", \"units\": \"dollars_per_call\"}]"},{"reference":"tests/test_sparse.py","line":260,"excerpt":"def test_daily_rules_are_not_silently_reused_from_hourly_rules():"},{"reference":"tests/test_sparse.py","line":279,"excerpt":"def test_adaptive_gap_does_not_silently_train_a_post_gap_episode():"},{"reference":"tests/test_sparse.py","line":322,"excerpt":"def test_restart_policy_cannot_reuse_a_prior_regime_reference():"},{"reference":"tests/test_sparse.py","line":431,"excerpt":"assert row[\"complete\"] and row[\"aggregate_value\"] == 24 and not row[\"model_eligible\"]"},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":220,"excerpt":"relationship cannot use unordered tied events. Completed-day event-index"},{"reference":"docs/sparse-data.md","line":239,"excerpt":"It then collects `calibration_size` eligible forecast residuals, not elapsed slots."},{"reference":"docs/sparse-data.md","line":261,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":332,"excerpt":"partial sums are progress metadata, never a completed total. Empty days are not"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":130,"excerpt":"source_grid_indexes=indexes, observed_sum=(math.fsum(measurements[i] for i in indexes if measurements[i] is not None)"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/agent.py","line":34,"excerpt":"def analyze_relationships(request: RequestV11 | dict) -> ResultV11:"},{"reference":"tests/test_sparse.py","line":10,"excerpt":"from anomalyzer import analyze, analyze_series, analyze_relationships, get_case, replay_policy"},{"reference":"tests/test_sparse.py","line":41,"excerpt":"assert right.data_quality[\"samples\"][8][\"source_index\"] == -1"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":61,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, 0, 0, 2]"},{"reference":"tests/test_sparse.py","line":63,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, None, 0, 2]"},{"reference":"docs/sparse-data.md","line":22,"excerpt":"works through `analyze`, `analyze_relationships`, CLI JSON and read-only MCP."},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":77,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":107,"excerpt":"retains its unique-source requirement. With an explicit `data_policy`, complete,"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","order":30}},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","status":"planned","area":"analysis","path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","health":"planned","coverage":"uncovered","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","order":10}},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","status":"planned","area":"analysis","path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","health":"planned","coverage":"uncovered","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","order":20}},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","status":"planned","area":"data","path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","health":"planned","coverage":"uncovered","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","order":10}},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","status":"planned","area":"data","path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","health":"planned","coverage":"uncovered","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","order":10}},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","status":"planned","area":"data","path":".capabilities/data/quality/report-source-freshness.capability.yaml","health":"planned","coverage":"uncovered","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","order":20}},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","status":"planned","area":"data","path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","order":30}},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","status":"planned","area":"evaluation","path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","order":30}},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","status":"planned","area":"evaluation","path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","health":"planned","coverage":"uncovered","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","order":20}},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","status":"planned","area":"evaluation","path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","order":10}},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","status":"planned","area":"execution","path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","order":20}},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","status":"planned","area":"execution","path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","order":10}},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","status":"planned","area":"interfaces","path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","order":20}},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","status":"planned","area":"investigation","path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","order":20}},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","order":10}},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","health":"planned","coverage":"uncovered","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","order":30}},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","status":"planned","area":"investigation","path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","order":30}},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","status":"planned","area":"relationships","path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","order":30}}],"byStoryMap":{"unassigned":[{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","status":"implemented","area":"data","path":".capabilities/data/quality/identify-distinct-events.capability.yaml","health":"review","coverage":"partial","summary":"Callers supply stable logical event IDs so distinct comparable calls at one timestamp retain separate values and evidence without inventing chronology.","intent":"Make real per-call evidence consumable while protecting references, replay and pattern interpretation from retry duplication and arbitrary tie order.","references":{"total":12,"readable":12,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Event-specific evidence and case identities survive deterministic replay; availability and identified corrections retain as-of provenance and bounded results.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":308,"excerpt":"identified = self.event_ids is not None"},{"reference":"src/anomalyzer/contracts.py","line":315,"excerpt":"raise ValueError(\"identified event revisions require event_id\")"},{"reference":"src/anomalyzer/contracts.py","line":319,"excerpt":"if trusted and (not identified or self.event_sequence is None):"},{"reference":"src/anomalyzer/contracts.py","line":596,"excerpt":"dataset_results: list[DatasetResultV11]"},{"reference":"src/anomalyzer/contracts.py","line":597,"excerpt":"relationship_results: list[RelationshipResult]"},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":14,"excerpt":"CONTRACT = \"event-identity-v1\""},{"reference":"src/anomalyzer/events.py","line":100,"excerpt":"key, identity = (revision.event_id, available), (revision.event_id, revision.revision_id)"},{"reference":"src/anomalyzer/events.py","line":102,"excerpt":"raise ValueError(\"duplicate event revision identity or availability\")"},{"reference":"src/anomalyzer/events.py","line":165,"excerpt":"arrival_semantics=\"supplied availability, otherwise event-time snapshot; no delivery latency claim\","},{"reference":"src/anomalyzer/sparse.py","line":182,"excerpt":"arrival_semantics=\"supplied availability timestamps; unannotated samples are event-time snapshots\" if any(s[\"available_at\"] and (s[\"value\"] is not None or s[\"coverage\"] == \"excluded\") for s in samples) else"},{"reference":"src/anomalyzer/sparse.py","line":194,"excerpt":"identified = \"event_identity_contract\" in quality"},{"reference":"src/anomalyzer/sparse.py","line":195,"excerpt":"resets = quality[\"event_reset_positions\"] if identified else active_resets(cfg.reset_points, times, ds.timezone, context.as_of)"},{"reference":"src/anomalyzer/sparse.py","line":202,"excerpt":"\"profile_revision\": \"identified-events-v1\" if identified else \"sparse-frozen-phase-v1\","},{"reference":"src/anomalyzer/sparse.py","line":333,"excerpt":"[method.model_copy(update={\"evidence\": [e]})], ds.id)) if identified else episodes([temporary], ds.id)"},{"reference":"src/anomalyzer/orchestrator.py","line":122,"excerpt":"dataset_results = []"},{"reference":"src/anomalyzer/orchestrator.py","line":143,"excerpt":"if \"event_identity_contract\" in source.quality else"},{"reference":"src/anomalyzer/orchestrator.py","line":155,"excerpt":"dataset_results.append(DatasetResultV11("},{"reference":"src/anomalyzer/orchestrator.py","line":162,"excerpt":"dataset_results.append(_empty_dataset_result(dataset.id, exc))"},{"reference":"src/anomalyzer/orchestrator.py","line":164,"excerpt":"relationship_results = []"},{"reference":"src/anomalyzer/alignment.py","line":63,"excerpt":"if any(\"event_identity_contract\" in source.quality for source in sources):"},{"reference":"src/anomalyzer/alignment.py","line":66,"excerpt":"if (not all(\"event_identity_contract\" in source.quality for source in sources)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/relationships.py","line":198,"excerpt":"event_order=policies[0].event_order if identified else \"timestamp\","},{"reference":"src/anomalyzer/relationships.py","line":203,"excerpt":"event_ids=[s[\"event_id\"] for s in sources[0].quality[\"samples\"]] if identified else None,"},{"reference":"src/anomalyzer/relationships.py","line":205,"excerpt":"event_sequence=[s[\"event_sequence\"] for s in sources[0].quality[\"samples\"]] if identified and policies[0].event_order == \"trusted_sequence\" else None)"},{"reference":"src/anomalyzer/cases.py","line":29,"excerpt":"def compose_cases(dataset_results: list[DatasetResultV11],"},{"reference":"src/anomalyzer/cases.py","line":30,"excerpt":"relationship_results: list[RelationshipResult],"},{"reference":"src/anomalyzer/cases.py","line":33,"excerpt":"for result in dataset_results:"},{"reference":"src/anomalyzer/cases.py","line":37,"excerpt":"for result in relationship_results:"},{"reference":"src/anomalyzer/cases.py","line":62,"excerpt":"case_id = \"case:\" + (case_digest if event_id is not None else case_digest[:16])"},{"reference":"src/anomalyzer/mcp_views.py","line":12,"excerpt":"for item in result[\"dataset_results\"]:"},{"reference":"src/anomalyzer/mcp_views.py","line":14,"excerpt":"for item in result[\"relationship_results\"]:"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":77,"excerpt":"if \"event_identity_contract\" in source else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":88,"excerpt":"def _case_assessments(result, case):"},{"reference":"tests/test_events.py","line":1,"excerpt":"\"\"\"Logical event identity, honest tie semantics and deterministic consumer replay.\"\"\""},{"reference":"tests/test_events.py","line":12,"excerpt":"from anomalyzer import analyze, analyze_series, get_case, replay_policy"},{"reference":"tests/test_events.py","line":36,"excerpt":"return [item for method in result.dataset_results[0].methods for item in method.evidence]"},{"reference":"tests/test_events.py","line":49,"excerpt":"item = result.dataset_results[0]"},{"reference":"tests/test_events.py","line":51,"excerpt":"assert item.data_quality[\"event_identity_contract\"] == \"event-identity-v1\""},{"reference":"docs/sparse-data.md","line":37,"excerpt":"print(result.dataset_results[0].data_quality[\"samples\"])"},{"reference":"docs/sparse-data.md","line":97,"excerpt":"Full evidence can be several megabytes for a bounded two-source history; the"},{"reference":"docs/sparse-data.md","line":100,"excerpt":"and result bytes for 1,344-hour and bounded event workloads. CI runs it in the"},{"reference":"docs/sparse-data.md","line":111,"excerpt":"rejected. Event-index timestamps without `event_ids` must be unique; identified"},{"reference":"docs/sparse-data.md","line":115,"excerpt":"## Stable event identities since 0.3.0"},{"reference":"examples/sparse_examples.py","line":39,"excerpt":"result[\"cost_events\"] = request(events)"},{"reference":"examples/sparse_examples.py","line":72,"excerpt":"result[\"identified_calls\"] = request(same_time, training_size=2, calibration_size=3)"},{"reference":"examples/sparse_examples.py","line":73,"excerpt":"corrected = deepcopy(result[\"identified_calls\"])"},{"reference":"examples/sparse_examples.py","line":77,"excerpt":"result[\"identified_corrected\"] = corrected"},{"reference":"examples/sparse_examples.py","line":80,"excerpt":"result[\"identified_before_correction\"] = before"},{"reference":"examples/render_readme_progression.py","line":3,"excerpt":"The figures use checked-in synthetic examples and real ``analyze`` results."},{"reference":"examples/render_readme_progression.py","line":781,"excerpt":"source_results = {item.dataset_id: item for item in result.dataset_results}"},{"reference":"examples/render_readme_progression.py","line":782,"excerpt":"relationship = result.relationship_results[0]"},{"reference":"examples/render_readme_progression.py","line":787,"excerpt":"assert not any(row[\"triggers\"] for item in result.dataset_results"},{"reference":"examples/render_readme_progression.py","line":798,"excerpt":"for row in source_results[dataset_id].methods[0].evidence}"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Explicit trusted sequence metadata establishes a validated event order, not independence, elapsed cadence or operational authority.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":246,"excerpt":"event_order: Literal[\"timestamp\", \"trusted_sequence\"] = \"timestamp\""},{"reference":"src/anomalyzer/contracts.py","line":255,"excerpt":"if self.event_order == \"trusted_sequence\" and self.observation_basis != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":256,"excerpt":"raise ValueError(\"trusted_sequence requires event_index\")"},{"reference":"src/anomalyzer/contracts.py","line":268,"excerpt":"EventId = Annotated[str, Field(min_length=1, max_length=128, strict=True, pattern=r\"^\\S(?:.*\\S)?$\")]"},{"reference":"src/anomalyzer/contracts.py","line":296,"excerpt":"event_sequence: list[Annotated[int, Field(ge=0, le=9223372036854775807, strict=True)]] | None = Field(default=None, json_schema_extra={\"uniqueItems\": True})"},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":50,"excerpt":"if policy.event_order != \"trusted_sequence\" and times[index] == times[index-1]:"},{"reference":"src/anomalyzer/events.py","line":71,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/events.py","line":80,"excerpt":"event_sequence=ds.event_sequence[index] if ds.event_sequence is not None else None,"},{"reference":"src/anomalyzer/events.py","line":84,"excerpt":"if policy.event_order == \"trusted_sequence\":"},{"reference":"src/anomalyzer/sparse.py","line":46,"excerpt":"if ds.event_ids is not None:"},{"reference":"src/anomalyzer/sparse.py","line":62,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/sparse.py","line":93,"excerpt":"raise ValueError(\"revision available_at cannot precede event time\")"},{"reference":"src/anomalyzer/sparse.py","line":170,"excerpt":"regularity_basis=\"declared_cadence\" if policy.observation_basis == \"calendar\" else \"explicit_comparable_events\","},{"reference":"src/anomalyzer/sparse.py","line":182,"excerpt":"arrival_semantics=\"supplied availability timestamps; unannotated samples are event-time snapshots\" if any(s[\"available_at\"] and (s[\"value\"] is not None or s[\"coverage\"] == \"excluded\") for s in samples) else"},{"reference":"src/anomalyzer/orchestrator.py","line":24,"excerpt":"from .events import annotate_assessments"},{"reference":"src/anomalyzer/orchestrator.py","line":142,"excerpt":"reset_positions=(source.quality[\"event_reset_positions\"] + source.quality[\"sequence_boundaries\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":193,"excerpt":"reset_positions=(result.data_quality[\"event_reset_positions\"] + result.data_quality[\"sequence_boundaries\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":209,"excerpt":"if assessment.event_id is not None:"},{"reference":"src/anomalyzer/orchestrator.py","line":213,"excerpt":"\"samples\", \"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")})"},{"reference":"src/anomalyzer/alignment.py","line":64,"excerpt":"signatures = [[(s.get(\"event_id\"), s.get(\"event_sequence\"))"},{"reference":"src/anomalyzer/alignment.py","line":66,"excerpt":"if (not all(\"event_identity_contract\" in source.quality for source in sources)"},{"reference":"src/anomalyzer/alignment.py","line":68,"excerpt":"or len({source.quality[\"event_order\"] for source in sources}) != 1):"},{"reference":"src/anomalyzer/relationships.py","line":38,"excerpt":"raise ValueError(\"lagged_response cannot order tied events without trusted_sequence\")"},{"reference":"src/anomalyzer/relationships.py","line":198,"excerpt":"event_order=policies[0].event_order if identified else \"timestamp\","},{"reference":"src/anomalyzer/relationships.py","line":205,"excerpt":"event_sequence=[s[\"event_sequence\"] for s in sources[0].quality[\"samples\"]] if identified and policies[0].event_order == \"trusted_sequence\" else None)"},{"reference":"src/anomalyzer/cases.py","line":59,"excerpt":"\"relationships\": relationships, **({\"event_id\": event_id} if event_id is not None else {})},"},{"reference":"src/anomalyzer/cases.py","line":62,"excerpt":"case_id = \"case:\" + (case_digest if event_id is not None else case_digest[:16])"},{"reference":"src/anomalyzer/cases.py","line":115,"excerpt":"for _, _, assessment in sorted(items, key=lambda entry: entry[2].id)]} if event_id is not None else {}),"},{"reference":"src/anomalyzer/cases.py","line":118,"excerpt":"revision_id = case_id + \":\" + (revision_digest if event_id is not None else revision_digest[:12])"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":152,"excerpt":"**({\"event_id\": case[\"event_id\"]} if case.get(\"event_id\") is not None else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":380,"excerpt":"if point.get(\"event_id\") is not None and point[\"index\"] in ("},{"reference":"src/anomalyzer/mcp_views.py","line":385,"excerpt":"\"note\": \"Evidence establishes numerical criteria only; source context, cause, impact, and action authority require external review.\","},{"reference":"tests/test_events.py","line":29,"excerpt":"ds[\"event_sequence\"] = list(range(100, 100+len(values)))"},{"reference":"tests/test_events.py","line":30,"excerpt":"ds[\"data_policy\"][\"event_order\"] = \"trusted_sequence\""},{"reference":"tests/test_events.py","line":80,"excerpt":"lambda d: d.update(event_sequence=list(range(7))),"},{"reference":"tests/test_events.py","line":81,"excerpt":"lambda d: d[\"data_policy\"].update(event_order=\"trusted_sequence\"),"},{"reference":"tests/test_events.py","line":121,"excerpt":"def test_trusted_sequence_is_explicit_and_enables_order_without_changing_time():"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":136,"excerpt":"| `event_ids` | Optional array aligned with original values/timestamps; every ID is a unique, case-sensitive string of 1–128 characters with no surrounding whitespace or line breaks. IDs require explicit event-index policy, original timestamps and no frequency. |"},{"reference":"docs/sparse-data.md","line":137,"excerpt":"| `data_policy.event_order` | `\"timestamp\"` by default. Sort by normalized time then exact ID for deterministic layout; equal-time ID order is **not** temporal order. |"},{"reference":"docs/sparse-data.md","line":138,"excerpt":"| `event_sequence` | Optional aligned array of unique strict integers in 0…9223372036854775807. Requires `event_order=\"trusted_sequence\"` and IDs. Ordinals must agree with nondecreasing original timestamps. Gaps in ordinals are allowed and do not represent missing events or elapsed time. |"},{"reference":"docs/sparse-data.md","line":139,"excerpt":"| `SampleRevision.event_id` | Required for every revision of an identified source. It addresses an existing ID at exactly the same original timestamp. A revision changes value/coverage, never event identity, time or sequence. |"},{"reference":"examples/sparse_examples.py","line":87,"excerpt":"ordered[\"event_sequence\"] = list(range(100,119))"},{"reference":"examples/sparse_examples.py","line":88,"excerpt":"ordered[\"data_policy\"][\"event_order\"] = \"trusted_sequence\""},{"reference":"examples/sparse_examples.py","line":89,"excerpt":"result[\"identified_pattern_sequence\"] = request(ordered, training_size=2, calibration_size=3, point_threshold=100)"},{"reference":"examples/render_readme_progression.py","line":918,"excerpt":"header(svg, \"17\", \"Compare real calls across quiet hours\", \"A per-call value is undefined in a no-call hour. Event position is an explicitly different observation basis.\")"},{"reference":"examples/render_readme_progression.py","line":921,"excerpt":"svg.text(95, 585, \"Event 0: Jan 1 00:00 UTC  ·  event 11: Jan 18 22:00 UTC  ·  no elapsed-time rate inferred\", 16, MUTED)"},{"reference":"examples/render_readme_progression.py","line":1024,"excerpt":"(\"identified_pattern_sequence\",410,\"Trusted sequence: ordered residual checks become applicable\")):"},{"reference":"examples/render_readme_progression.py","line":1037,"excerpt":"svg.text(60,665,\"Only assert trusted_sequence from actual source order. Neither sequence nor distinct IDs proves independence or cause.\",15,MUTED)"},{"reference":"examples/render_readme_progression.py","line":1038,"excerpt":"figures[\"anomaly-progression-24-event-order\"] = svg.finish()"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/identify-distinct-events","title":"Inspect distinct events sharing an original timestamp","criterion":"Identified event-index sources retain every distinct event and its original time, reject repeated logical IDs and preserve calendar and legacy semantics.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":118,"excerpt":"score_semantics: str"},{"reference":"src/anomalyzer/contracts.py","line":192,"excerpt":"# This avoids changing the generated 1.0 schemas while giving agent-facing"},{"reference":"src/anomalyzer/contracts.py","line":193,"excerpt":"# callers explicit evidence and relationship semantics."},{"reference":"src/anomalyzer/contracts.py","line":241,"excerpt":"\"\"\"Explicit source semantics; unknown collection is never implicit zero.\"\"\""},{"reference":"src/anomalyzer/events.py","line":1,"excerpt":"\"\"\"event-identity-v1: stable logical identity, original time, and explicit order."},{"reference":"src/anomalyzer/events.py","line":14,"excerpt":"CONTRACT = \"event-identity-v1\""},{"reference":"src/anomalyzer/events.py","line":22,"excerpt":"def event_token(event_id, entity):"},{"reference":"src/anomalyzer/events.py","line":23,"excerpt":"return fingerprint(dict(event_id=event_id, entity=entity))"},{"reference":"src/anomalyzer/events.py","line":26,"excerpt":"def event_resets(points, times, policy, zone, as_of):"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":23,"excerpt":"from .validation import resolve_reset_points, timestamp"},{"reference":"src/anomalyzer/sparse.py","line":46,"excerpt":"if ds.event_ids is not None:"},{"reference":"src/anomalyzer/sparse.py","line":60,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":62,"excerpt":"raise ValueError(\"available_at cannot precede event time\")"},{"reference":"src/anomalyzer/orchestrator.py","line":21,"excerpt":"from .relationships import derive, _source_ids"},{"reference":"src/anomalyzer/orchestrator.py","line":84,"excerpt":"evidence_ids = {}"},{"reference":"src/anomalyzer/orchestrator.py","line":90,"excerpt":"evidence_ids[old] = item[\"id\"]"},{"reference":"src/anomalyzer/orchestrator.py","line":94,"excerpt":"observation[\"evidence_refs\"] = [evidence_ids.get(value, value)"},{"reference":"src/anomalyzer/orchestrator.py","line":100,"excerpt":"pattern[\"evidence_refs\"] = [evidence_ids.get(value, value)"},{"reference":"src/anomalyzer/alignment.py","line":14,"excerpt":"source_indexes: list[int]"},{"reference":"src/anomalyzer/alignment.py","line":33,"excerpt":"indexes = ([sample[\"source_index\"] if len(sample[\"source_indexes\"]) <= 1 else -1"},{"reference":"src/anomalyzer/alignment.py","line":46,"excerpt":"def exact_sources(ids: list[str], prepared: dict[str, PreparedDataset]):"},{"reference":"src/anomalyzer/alignment.py","line":47,"excerpt":"sources = [prepared[source_id] for source_id in ids]"},{"reference":"src/anomalyzer/alignment.py","line":48,"excerpt":"if any(source.dataset.timestamps is None for source in sources):"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/cases.py","line":16,"excerpt":"source_id = relationship.numerator"},{"reference":"src/anomalyzer/cases.py","line":18,"excerpt":"source_id = relationship.minuend"},{"reference":"src/anomalyzer/cases.py","line":20,"excerpt":"source_id = relationship.observed"},{"reference":"src/anomalyzer/cases.py","line":22,"excerpt":"source_id = relationship.response"},{"reference":"src/anomalyzer/cases.py","line":24,"excerpt":"source_id = relationship.conditions[0].dataset"},{"reference":"src/anomalyzer/mcp_views.py","line":68,"excerpt":"\"latest_evidence_state\": source.get(\"latest_evidence_state\"),"},{"reference":"src/anomalyzer/mcp_views.py","line":71,"excerpt":"\"grid_count\": source[\"grid_count\"], \"usable_model_count\": source[\"usable_model_count\"],"},{"reference":"src/anomalyzer/mcp_views.py","line":76,"excerpt":"**({key: source[key] for key in (\"event_identity_contract\", \"event_order\", \"tied_timestamp_count\", \"tied_event_count\")}"},{"reference":"src/anomalyzer/mcp_views.py","line":77,"excerpt":"if \"event_identity_contract\" in source else {}),"},{"reference":"src/anomalyzer/mcp_views.py","line":89,"excerpt":"ids = set(case[\"supporting_evidence\"])"},{"reference":"tests/test_events.py","line":1,"excerpt":"\"\"\"Logical event identity, honest tie semantics and deterministic consumer replay.\"\"\""},{"reference":"tests/test_events.py","line":22,"excerpt":"ds = dict(id=\"cost\", entity={\"application\": \"synthetic\"}, units=\"dollars_per_call\","},{"reference":"tests/test_events.py","line":24,"excerpt":"event_ids=[f\"call-{i:04}\" for i in range(len(values))],"},{"reference":"tests/test_events.py","line":25,"excerpt":"data_policy={\"observation_basis\": \"event_index\"},"},{"reference":"tests/test_events.py","line":41,"excerpt":"return {k: semantic(v) for k, v in value.items() if k not in (\"run_id\", \"runtime_seconds\", \"source_index\", \"source_indexes\")}"},{"reference":"docs/sparse-data.md","line":50,"excerpt":"`coverage`, `available_at` and legacy `period_statuses` align with the original"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":74,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":77,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":83,"excerpt":"## Calendar coordinates and bounded grids"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":65,"excerpt":"# Genuine distinct calls share the recorded timestamp; IDs do not order them."},{"reference":"examples/sparse_examples.py","line":66,"excerpt":"same_time = {\"id\":\"call-cost\", \"entity\":{\"application\":\"synthetic\"}, \"units\":\"dollars_per_call\","},{"reference":"examples/sparse_examples.py","line":69,"excerpt":"\"event_ids\":[f\"call-{i:02}\" for i in range(10)],"},{"reference":"examples/render_readme_progression.py","line":141,"excerpt":"sx = lambda index: x + index / max(count - 1, 1) * width"},{"reference":"examples/render_readme_progression.py","line":301,"excerpt":"frozen_flags = {row[\"index\"] for row in frozen.methods[0].evidence"},{"reference":"examples/render_readme_progression.py","line":303,"excerpt":"adaptive_flags = {row[\"index\"] for row in adaptive.methods[0].evidence"},{"reference":"examples/render_readme_progression.py","line":373,"excerpt":"sx = lambda index: x + index / (len(sample) - 1) * width"},{"reference":"examples/render_readme_progression.py","line":413,"excerpt":"scores = {row[\"index\"]: row[\"standardized_residual\"] for row in evidence}"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}]}],"releases":[{"release":"Anomaly detection Skill","capabilities":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/guide-anomaly-investigation.capability.yaml","health":"review","coverage":"partial","summary":"The reusable anomaly-detection Skill guides an agent through local Anomalyzer analysis and a concise explanation grounded in structured evidence.","intent":"Make the analysis usable through an agent while keeping data handling and interpretation limits explicit.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The default response instructions lead with changes and observed-versus-expected values, then explain readiness, evidence maturity, overlapping findings, and material limitations without inventing confidence percentages or causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill describes MCP analyze, list_findings, and get_evidence as an optional evidence workflow and makes session lifetime, read-only access, and caller-owned notification and action delivery explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/guide-anomaly-investigation","title":"Ask an agent to explain unusual changes in my metrics","criterion":"The Skill directs analysis through the installed Anomalyzer CLI, with executable fallback paths and structured JSON output, rather than a newly invented detector.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/guide-anomaly-investigation --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","order":10}},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","status":"implemented","area":"interfaces","path":".capabilities/interfaces/skill/visualize-analysis-on-request.capability.yaml","health":"review","coverage":"partial","summary":"The anomaly-detection Skill offers an explicitly requested quick chart or a detailed figure specification after structured analysis completes.","intent":"Make findings easier to inspect visually while preserving their evidence meaning and the user's requested level of detail.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"Detailed figure instructions preserve source indices, provisional-versus-calibrated evidence, pattern detection timing, and incomplete or failed run status, without presenting scores as probabilities or flags as proven causes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The dependency-free quick chart fallback renders one resolved dataset's observed values and available expectations, marks only triggered evidence, and labels training, calibration, units, and reviewed reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/skill/visualize-analysis-on-request","title":"See the evidence in a chart when I ask for one","criterion":"The fallback rejects unsupported multi-dataset results and protects an existing output file unless overwrite is explicitly requested.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/skill/visualize-analysis-on-request --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Anomaly detection Skill","backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","order":20}}],"deliveryStrategy":{"release":"Anomaly detection Skill","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/skill/guide-anomaly-investigation"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/skill/guide-anomaly-investigation needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Ask an agent to investigate a change"],"missingBackbones":[],"stepCoverageGaps":["Ask an agent to investigate a change > Inspect a requested chart"],"rationale":"Starts with the earliest mapped step in Ask an agent to investigate a change."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/skill/visualize-analysis-on-request"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/skill/visualize-analysis-on-request needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Ask an agent to investigate a change"],"missingBackbones":[],"stepCoverageGaps":["Ask an agent to investigate a change > Get an evidence-grounded explanation"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/skill/visualize-analysis-on-request."}]},"presentation":{"outcome":"Anomaly detection Skill helps teams move from Get an evidence-grounded explanation to Inspect a requested chart across 1 backbone activity.","narrativePath":[{"backbone":"Ask an agent to investigate a change","step":"Get an evidence-grounded explanation","capabilityIds":["interfaces/skill/guide-anomaly-investigation"],"health":"review"},{"backbone":"Ask an agent to investigate a change","step":"Inspect a requested chart","capabilityIds":["interfaces/skill/visualize-analysis-on-request"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Ask an agent to investigate a change > Get an evidence-grounded explanation","message":"Ask an agent to investigate a change > Get an evidence-grounded explanation has needs-review coverage."},{"kind":"weak","label":"Ask an agent to investigate a change > Inspect a requested chart","message":"Ask an agent to investigate a change > Inspect a requested chart has needs-review coverage."}]}},{"release":"Basic CLI anomaly detection","capabilities":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/report-point-anomaly-episodes.capability.yaml","health":"review","coverage":"partial","summary":"Calibrated point-threshold crossings become inspectable observations, with adjacent crossings grouped into episodes.","intent":"Locate unusual values and keep isolated events distinct from sustained runs.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"A sufficiently extreme calibrated sample records a point trigger and the expectation it departed from.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Adjacent point departures form a consecutive-run pattern with the contributing sample positions.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/report-point-anomaly-episodes","title":"Find unusual spikes and drops","criterion":"Repeated runs of the same fixed request preserve numerical evidence; future samples do not change earlier scoring for declared settings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/report-point-anomaly-episodes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Find spikes and drops","order":40}},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/calibrate-departures.capability.yaml","health":"review","coverage":"partial","summary":"Sample evidence distinguishes reference-only, provisional, and calibrated comparisons before permitting statistical point triggers.","intent":"Keep useful early evidence while making its level of support visible.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Completed calibration provides a frozen residual center and sample scale for later standardized residuals.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Evidence records the observed value, expectation, calibration support, and numerical criterion responsible for a trigger.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/calibrate-departures","title":"Understand how much history supports a finding","criterion":"Provisional residual scores can be reported but do not trigger point-anomaly episodes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/calibrate-departures --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Establish evidence support","order":30}},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","status":"implemented","area":"data","path":".capabilities/data/inputs/accept-analysis-requests.capability.yaml","health":"review","coverage":"partial","summary":"Callers can submit ordered numeric values or versioned requests containing datasets, settings, context, and declared relationships.","intent":"Make the accepted data contract clear before any conclusions are drawn from the numbers.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Ordered value lists work with separate settings, while schema 1.0 requests describe one dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Published schema 1.0 and 1.1 request and result schemas match the public Python contracts.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/inputs/accept-analysis-requests","title":"Bring my metrics into an analysis","criterion":"Schema 1.1 datasets declare a nonempty entity scope and support explicit relationships between named datasets.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/inputs/accept-analysis-requests --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Supply a numeric series","order":10}},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","status":"implemented","area":"data","path":".capabilities/data/quality/prepare-timestamped-series.capability.yaml","health":"review","coverage":"partial","summary":"Timestamped observations are normalized to UTC, ordered, and checked against their declared cadence before analysis.","intent":"Compare like periods and make preparation decisions visible to the caller.","references":{"total":9,"readable":9,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"An as_of cutoff excludes later observations, and cadence is checked as a fixed elapsed duration.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar datasets provide timestamps and frequency; positional datasets provide neither. Explicit schema-1.1 event-index sources retain timestamps without a calendar frequency.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/prepare-timestamped-series","title":"Analyze observations in the right time order","criterion":"Calendar duplicate timestamps require an aggregation policy; identified event-index calls can share original timestamps while duplicate logical IDs are rejected and arbitrary tie order supplies no sequence evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/prepare-timestamped-series --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Prepare the comparison","order":20}},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/analyze-files-and-streams.capability.yaml","health":"review","coverage":"partial","summary":"The CLI analyzes CSV, JSON requests, and bare JSON values from files or stdin, with configurable settings and protected output files.","intent":"Make local analysis usable in scripts and ad hoc investigation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"CSV supports positional or timestamped selected values, treats empty cells as missing, and rejects varying unmapped entity columns.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Equivalent file and stdin JSON inputs produce equivalent request fingerprints and analysis semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/analyze-files-and-streams","title":"Analyze local files from the command line","criterion":"Output files require explicit overwrite permission, and invalid input is distinguished from execution failure by exit status.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/analyze-files-and-streams --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Run from files or stdin","order":50}},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","status":"implemented","area":"interfaces","path":".capabilities/interfaces/cli/explain-findings-in-readable-output.capability.yaml","health":"review","coverage":"partial","summary":"Readable CLI output distinguishes point episodes, pattern findings, cases, explicit rule violations, and reference readiness.","intent":"Let a person understand what changed and which evidence supports it without parsing the full JSON result.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Explicit numerical rules are named with their boundaries instead of appearing as statistical anomaly claims.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Point-anomaly episode counts and residual-pattern findings are labeled separately.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/cli/explain-findings-in-readable-output","title":"Understand findings in a readable summary","criterion":"Readiness status remains visible, including when it was not assessed.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/cli/explain-findings-in-readable-output --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Basic CLI anomaly detection","backbone":"Analyze a local series","step":"Read the findings","order":60}}],"deliveryStrategy":{"release":"Basic CLI anomaly detection","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/inputs/accept-analysis-requests"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/inputs/accept-analysis-requests needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Analyze a local series"],"missingBackbones":[],"stepCoverageGaps":["Analyze a local series > Establish evidence support","Analyze a local series > Find spikes and drops","Analyze a local series > Prepare the comparison","Analyze a local series > Read the findings","Analyze a local series > Run from files or stdin"],"rationale":"Starts with the earliest mapped step in Analyze a local series."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["data/quality/prepare-timestamped-series","analysis/evidence/calibrate-departures","analysis/detection/report-point-anomaly-episodes","interfaces/cli/analyze-files-and-streams","interfaces/cli/explain-findings-in-readable-output"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/detection/report-point-anomaly-episodes needs review; analysis/evidence/calibrate-departures needs review; data/quality/prepare-timestamped-series needs review; interfaces/cli/analyze-files-and-streams needs review; interfaces/cli/explain-findings-in-readable-output needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Analyze a local series"],"missingBackbones":[],"stepCoverageGaps":["Analyze a local series > Supply a numeric series"],"rationale":"Prioritizes capabilities with delivery risk signals: data/quality/prepare-timestamped-series, analysis/evidence/calibrate-departures, analysis/detection/report-point-anomaly-episodes, interfaces/cli/analyze-files-and-streams, interfaces/cli/explain-findings-in-readable-output."}]},"presentation":{"outcome":"Basic CLI anomaly detection helps teams move from Supply a numeric series to Read the findings across 1 backbone activity.","narrativePath":[{"backbone":"Analyze a local series","step":"Supply a numeric series","capabilityIds":["data/inputs/accept-analysis-requests"],"health":"review"},{"backbone":"Analyze a local series","step":"Prepare the comparison","capabilityIds":["data/quality/prepare-timestamped-series"],"health":"review"},{"backbone":"Analyze a local series","step":"Establish evidence support","capabilityIds":["analysis/evidence/calibrate-departures"],"health":"review"},{"backbone":"Analyze a local series","step":"Find spikes and drops","capabilityIds":["analysis/detection/report-point-anomaly-episodes"],"health":"review"},{"backbone":"Analyze a local series","step":"Run from files or stdin","capabilityIds":["interfaces/cli/analyze-files-and-streams"],"health":"review"},{"backbone":"Analyze a local series","step":"Read the findings","capabilityIds":["interfaces/cli/explain-findings-in-readable-output"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Analyze a local series > Establish evidence support","message":"Analyze a local series > Establish evidence support has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Find spikes and drops","message":"Analyze a local series > Find spikes and drops has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Prepare the comparison","message":"Analyze a local series > Prepare the comparison has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Read the findings","message":"Analyze a local series > Read the findings has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Run from files or stdin","message":"Analyze a local series > Run from files or stdin has needs-review coverage."},{"kind":"weak","label":"Analyze a local series > Supply a numeric series","message":"Analyze a local series > Supply a numeric series has needs-review coverage."}]}},{"release":"Evidence and review policy","capabilities":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","status":"implemented","area":"analysis","path":".capabilities/analysis/evidence/assess-reference-readiness.capability.yaml","health":"review","coverage":"partial","summary":"Readiness diagnostics report supported, caution, or not_assessed based on the training pattern and calibration evidence.","intent":"Expose weak expectations even when no anomaly is detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Insufficient calibration support yields not_assessed, including when the run otherwise completes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Later evaluation changes do not rewrite the segment's training-and-calibration readiness assessment.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/evidence/assess-reference-readiness","title":"Know when the expected baseline is weak","criterion":"Repeating seasonal structure with stable calibration can yield supported readiness.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/evidence/assess-reference-readiness --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","order":20}},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","status":"implemented","area":"execution","path":".capabilities/execution/resilience/retain-evidence-under-failure-and-limits.capability.yaml","health":"review","coverage":"partial","summary":"Analysis applies cooperative resource limits and exposes partial or failed results without discarding valid earlier or sibling evidence.","intent":"Make constrained or failed runs understandable instead of silently returning incomplete conclusions.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A failed dataset or relationship leaves healthy sibling results available and marks the overall run partial.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"A method failure preserves evidence completed before the failure and reports the error.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"execution/resilience/retain-evidence-under-failure-and-limits","title":"Keep useful evidence when analysis cannot finish","criterion":"CLI byte limits and execution failures produce distinct error exits.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify execution/resilience/retain-evidence-under-failure-and-limits --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","order":50}},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","status":"implemented","area":"investigation","path":".capabilities/investigation/assessments/describe-early-changes.capability.yaml","health":"review","coverage":"partial","summary":"Short histories yield observations, changes, and departure candidates with explicit comparison populations and limits.","intent":"Provide an inspectable lead for investigation without overstating statistical support.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Descriptive assessments use preceding observations and remain unchanged when future samples are appended.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Early candidates remain non-triggering and ineligible for action even under a permissive policy.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/assessments/describe-early-changes","title":"Spot early changes with a short history","criterion":"Missing values break the comparison history, and zero previous values do not create relative-change evidence.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/assessments/describe-early-changes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","order":10}},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","status":"implemented","area":"investigation","path":".capabilities/investigation/cases/group-related-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Qualifying assessments are grouped by event time and entity into cases with explanations, evidence references, and numerical severity.","intent":"Give investigators a coherent account of related evidence without counting derived metrics as independent confirmation.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"A derived relationship adds context without counting as an additional independent source for severity.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Adaptive location-shift rules can create an early grouped case without a point trigger, using only evidence available at detection.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/cases/group-related-evidence","title":"Investigate related findings together","criterion":"Criteria met in multiple source datasets can raise numerical severity and identify the contributing sources.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/cases/group-related-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","order":40}},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","status":"implemented","area":"investigation","path":".capabilities/investigation/policy/replay-action-eligibility.capability.yaml","health":"review","coverage":"partial","summary":"Callers can apply a different explicit action policy to existing assessments and cases without recomputing numerical analysis.","intent":"Separate an observed criterion from a caller's decision to make it eligible for further action.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Default policy leaves action and notification eligibility disabled.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Explicit-rule and calibrated-departure eligibility are controlled independently.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/policy/replay-action-eligibility","title":"Change action eligibility without rerunning analysis","criterion":"Notification eligibility can be allowed for a met criterion without action eligibility, while early candidates remain ineligible.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/policy/replay-action-eligibility --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Review action eligibility","order":60}},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","status":"implemented","area":"investigation","path":".capabilities/investigation/rules/check-declared-numerical-rules.capability.yaml","health":"review","coverage":"partial","summary":"Acceptable ranges and maximum absolute or relative changes identify explicit criterion violations, including before calibration.","intent":"Let callers check clear numerical constraints without presenting them as learned anomaly thresholds.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Absolute and relative changes must exceed their declared maximum to violate a rule.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Relative-change checks require a nonzero previous value and do not bridge missing observations.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"investigation/rules/check-declared-numerical-rules","title":"Find values that break my stated limits","criterion":"Values outside an inclusive acceptable range are violations, while values on the boundary are accepted.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify investigation/rules/check-declared-numerical-rules --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Evidence and review policy","backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","order":30}}],"deliveryStrategy":{"release":"Evidence and review policy","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["investigation/assessments/describe-early-changes"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"investigation/assessments/describe-early-changes needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Decide how much to rely on a finding"],"missingBackbones":[],"stepCoverageGaps":["Decide how much to rely on a finding > Apply my stated limits","Decide how much to rely on a finding > Assess baseline quality","Decide how much to rely on a finding > Inspect incomplete execution","Decide how much to rely on a finding > Investigate related evidence","Decide how much to rely on a finding > Review action eligibility"],"rationale":"Starts with the earliest mapped step in Decide how much to rely on a finding."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/evidence/assess-reference-readiness","investigation/rules/check-declared-numerical-rules","investigation/cases/group-related-evidence","execution/resilience/retain-evidence-under-failure-and-limits","investigation/policy/replay-action-eligibility"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/evidence/assess-reference-readiness needs review; execution/resilience/retain-evidence-under-failure-and-limits needs review; investigation/cases/group-related-evidence needs review; investigation/policy/replay-action-eligibility needs review; investigation/rules/check-declared-numerical-rules needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Decide how much to rely on a finding"],"missingBackbones":[],"stepCoverageGaps":["Decide how much to rely on a finding > Inspect early evidence"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/evidence/assess-reference-readiness, investigation/rules/check-declared-numerical-rules, investigation/cases/group-related-evidence, execution/resilience/retain-evidence-under-failure-and-limits, investigation/policy/replay-action-eligibility."}]},"presentation":{"outcome":"Evidence and review policy helps teams move from Inspect early evidence to Review action eligibility across 1 backbone activity.","narrativePath":[{"backbone":"Decide how much to rely on a finding","step":"Inspect early evidence","capabilityIds":["investigation/assessments/describe-early-changes"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Assess baseline quality","capabilityIds":["analysis/evidence/assess-reference-readiness"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Apply my stated limits","capabilityIds":["investigation/rules/check-declared-numerical-rules"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Investigate related evidence","capabilityIds":["investigation/cases/group-related-evidence"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Inspect incomplete execution","capabilityIds":["execution/resilience/retain-evidence-under-failure-and-limits"],"health":"review"},{"backbone":"Decide how much to rely on a finding","step":"Review action eligibility","capabilityIds":["investigation/policy/replay-action-eligibility"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Decide how much to rely on a finding > Apply my stated limits","message":"Decide how much to rely on a finding > Apply my stated limits has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Assess baseline quality","message":"Decide how much to rely on a finding > Assess baseline quality has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Inspect early evidence","message":"Decide how much to rely on a finding > Inspect early evidence has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Inspect incomplete execution","message":"Decide how much to rely on a finding > Inspect incomplete execution has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Investigate related evidence","message":"Decide how much to rely on a finding > Investigate related evidence has needs-review coverage."},{"kind":"weak","label":"Decide how much to rely on a finding > Review action eligibility","message":"Decide how much to rely on a finding > Review action eligibility has needs-review coverage."}]}},{"release":"Future - Business-aware expectations","capabilities":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","status":"planned","area":"analysis","path":".capabilities/analysis/expectations/model-multiple-seasonal-cycles.capability.yaml","health":"planned","coverage":"uncovered","summary":"Build an inspectable expectation from a bounded set of caller-declared seasonal cycles instead of forcing one cycle to explain every repeating pattern.","intent":"Reduce departures caused by ordinary combinations such as hour-of-day and weekday behavior while retaining evidence of unusual changes.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated or manual verification checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/expectations/model-multiple-seasonal-cycles has no automated checks","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"},{"code":"missing-manual-review","message":"analysis/expectations/model-multiple-seasonal-cycles has no manual review guidance","capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\expectations\\model-multiple-seasonal-cycles.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Calibration remains separated from expectation fitting, and an isolated extreme sample retains evidence without silently contaminating every cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Callers explicitly declare supported seasonal cycles and their elapsed or calendar meaning; duplicate or incompatible cycle definitions are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."},{"capabilityId":"analysis/expectations/model-multiple-seasonal-cycles","title":"Account for daily and weekly patterns together","criterion":"Expectations use only earlier observations and expose the contribution and available history of each modeled cycle.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/expectations/model-multiple-seasonal-cycles, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","order":10}},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","status":"planned","area":"analysis","path":".capabilities/analysis/rates/assess-bounded-count-rates.capability.yaml","health":"planned","coverage":"uncovered","summary":"Analyze explicitly declared success-count and eligible-count rates with denominator-aware evidence and a stated count-model assumption.","intent":"Distinguish a sparse conversion swing from a well-supported change without losing the numerator and denominator behind the rate.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"analysis/rates/assess-bounded-count-rates has no automated or manual verification checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-automated-checks","message":"analysis/rates/assess-bounded-count-rates has no automated checks","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"},{"code":"missing-manual-review","message":"analysis/rates/assess-bounded-count-rates has no manual review guidance","capabilityId":"analysis/rates/assess-bounded-count-rates","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\analysis\\rates\\assess-bounded-count-rates.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Callers declare aligned success and eligible counts; non-integer, negative, or success-above-eligible inputs are rejected for the bounded-count model.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Evidence retains numerator, denominator, rate, and the declared reference population rather than reporting a percentage alone.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."},{"capabilityId":"analysis/rates/assess-bounded-count-rates","title":"Interpret rate changes with their supporting volume","criterion":"Statistical rate departures identify the model and support used and remain distinct from exact caller-supplied rate-limit violations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to analysis/rates/assess-bounded-count-rates, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Judge rates using their exposure","order":20}},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","status":"planned","area":"investigation","path":".capabilities/investigation/regimes/propose-reviewed-baseline-changes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Propose baseline-change candidates from sustained evidence and compare existing and proposed regimes before a caller explicitly accepts a reset.","intent":"Help distinguish an unresolved departure from a reviewed process change without automatically normalizing away an incident.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated or manual verification checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/regimes/propose-reviewed-baseline-changes has no automated checks","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/regimes/propose-reviewed-baseline-changes has no manual review guidance","capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\regimes\\propose-reviewed-baseline-changes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A preview shows old and proposed baselines, calibration support, and findings affected by the reset, including insufficient history after the boundary.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"A proposal records the candidate boundary, first detection time, supporting observations, and limitations; estimated onset is distinct from detection time.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/regimes/propose-reviewed-baseline-changes","title":"Review a possible new baseline before accepting it","criterion":"Only an explicit caller acceptance produces a new request with a declared reset; rejected proposals remain inspectable and have no numerical effect.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/regimes/propose-reviewed-baseline-changes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Business-aware expectations","backbone":"Make expectations fit the process","step":"Review a proposed regime transition","order":30}}],"deliveryStrategy":{"release":"Future - Business-aware expectations","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["analysis/expectations/model-multiple-seasonal-cycles"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"analysis/expectations/model-multiple-seasonal-cycles has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Make expectations fit the process"],"missingBackbones":[],"stepCoverageGaps":["Make expectations fit the process > Judge rates using their exposure","Make expectations fit the process > Review a proposed regime transition"],"rationale":"Starts with the earliest mapped step in Make expectations fit the process."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["analysis/rates/assess-bounded-count-rates","investigation/regimes/propose-reviewed-baseline-changes"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"analysis/rates/assess-bounded-count-rates has verification gaps; investigation/regimes/propose-reviewed-baseline-changes has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Make expectations fit the process"],"missingBackbones":[],"stepCoverageGaps":["Make expectations fit the process > Represent more than one repeating cycle"],"rationale":"Follows story-map order across Judge rates using their exposure, Review a proposed regime transition."}]},"presentation":{"outcome":"Future - Business-aware expectations helps teams move from Represent more than one repeating cycle to Review a proposed regime transition across 1 backbone activity.","narrativePath":[{"backbone":"Make expectations fit the process","step":"Represent more than one repeating cycle","capabilityIds":["analysis/expectations/model-multiple-seasonal-cycles"],"health":"planned"},{"backbone":"Make expectations fit the process","step":"Judge rates using their exposure","capabilityIds":["analysis/rates/assess-bounded-count-rates"],"health":"planned"},{"backbone":"Make expectations fit the process","step":"Review a proposed regime transition","capabilityIds":["investigation/regimes/propose-reviewed-baseline-changes"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Calendar and data readiness","capabilities":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","status":"planned","area":"data","path":".capabilities/data/calendar/aggregate-local-calendar-periods.capability.yaml","health":"planned","coverage":"uncovered","summary":"Aggregate completed local calendar days or weeks using an explicit timezone and report the actual boundaries and supporting observations.","intent":"Align business-period comparisons with the calendar people use, including daylight-saving transitions.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/calendar/aggregate-local-calendar-periods has no automated or manual verification checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-automated-checks","message":"data/calendar/aggregate-local-calendar-periods has no automated checks","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"},{"code":"missing-manual-review","message":"data/calendar/aggregate-local-calendar-periods has no manual review guidance","capabilityId":"data/calendar/aggregate-local-calendar-periods","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\calendar\\aggregate-local-calendar-periods.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Callers select local-day or local-week aggregation, an IANA timezone, a week-start rule when needed, and sum, mean, or last semantics.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Completeness is determined from the declared native cadence and source completeness; missing slots and unfinished periods remain explicit.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."},{"capabilityId":"data/calendar/aggregate-local-calendar-periods","title":"Compare days and weeks in my business timezone","criterion":"Existing fixed-duration analysis remains reproducible, and ambiguous local input timestamps still require caller disambiguation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/calendar/aggregate-local-calendar-periods, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","order":10}},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","status":"planned","area":"data","path":".capabilities/data/quality/report-source-freshness.capability.yaml","health":"planned","coverage":"uncovered","summary":"Evaluate caller-declared update deadlines and allowed lateness separately from value anomalies and propagate stale-source limitations to dependent evidence.","intent":"Prevent an old or absent observation from being interpreted as current healthy behavior or a confirmed business change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/quality/report-source-freshness has no automated or manual verification checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-automated-checks","message":"data/quality/report-source-freshness has no automated checks","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"},{"code":"missing-manual-review","message":"data/quality/report-source-freshness has no manual review guidance","capabilityId":"data/quality/report-source-freshness","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\quality\\report-source-freshness.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"A stale source leaves its historical numerical evidence available while marking dependent current relationship conclusions as stale or unavailable.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Callers supply an evaluation time, expected update schedule, timezone when relevant, and allowed lateness for each source.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."},{"capabilityId":"data/quality/report-source-freshness","title":"Know when missing updates weaken the conclusion","criterion":"Re-evaluating freshness at another supplied time preserves historical numerical evidence and records the changed freshness decision.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/quality/report-source-freshness, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Calendar and data readiness","backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","order":20}}],"deliveryStrategy":{"release":"Future - Calendar and data readiness","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/calendar/aggregate-local-calendar-periods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/calendar/aggregate-local-calendar-periods has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Monitor the periods my business uses > Check whether the sources are current"],"rationale":"Starts with the earliest mapped step in Monitor the periods my business uses."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["data/quality/report-source-freshness"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"data/quality/report-source-freshness has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Monitor the periods my business uses > Declare local calendar boundaries"],"rationale":"Follows story-map order across Check whether the sources are current."}]},"presentation":{"outcome":"Future - Calendar and data readiness helps teams move from Declare local calendar boundaries to Check whether the sources are current across 1 backbone activity.","narrativePath":[{"backbone":"Monitor the periods my business uses","step":"Declare local calendar boundaries","capabilityIds":["data/calendar/aggregate-local-calendar-periods"],"health":"planned"},{"backbone":"Monitor the periods my business uses","step":"Check whether the sources are current","capabilityIds":["data/quality/report-source-freshness"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Cohort investigation","capabilities":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","status":"planned","area":"data","path":".capabilities/data/cohorts/analyze-declared-entity-batches.capability.yaml","health":"planned","coverage":"uncovered","summary":"Run a bounded batch of explicitly partitioned entity series and expose per-entity findings, support, and failures in one reviewable result.","intent":"Help callers inspect customers, regions, or services without collapsing their distinct baselines into one misleading series.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/cohorts/analyze-declared-entity-batches has no automated or manual verification checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-automated-checks","message":"data/cohorts/analyze-declared-entity-batches has no automated checks","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"},{"code":"missing-manual-review","message":"data/cohorts/analyze-declared-entity-batches has no manual review guidance","capabilityId":"data/cohorts/analyze-declared-entity-batches","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\cohorts\\analyze-declared-entity-batches.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Callers declare entity keys, metric units, cadence, and a batch resource budget; each entity receives a separate validated analysis request.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"Results report the evaluated population and raw finding counts without implying that many simultaneous tests establish a confirmed incident.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."},{"capabilityId":"data/cohorts/analyze-declared-entity-batches","title":"Analyze many entities without mixing their histories","criterion":"The result preserves each entity's findings, insufficient-history state, data quality, and failure status; one failure leaves healthy entities available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/cohorts/analyze-declared-entity-batches, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","order":10}},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","status":"planned","area":"investigation","path":".capabilities/investigation/cohorts/explain-additive-segment-contributions.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose an aggregate departure into contributions from caller-declared disjoint additive segments with compatible periods and units.","intent":"Direct investigation toward the parts of a total that changed while keeping arithmetic contribution distinct from causal explanation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/cohorts/explain-additive-segment-contributions has no automated or manual verification checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/cohorts/explain-additive-segment-contributions has no automated checks","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"},{"code":"missing-manual-review","message":"investigation/cohorts/explain-additive-segment-contributions has no manual review guidance","capabilityId":"investigation/cohorts/explain-additive-segment-contributions","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\cohorts\\explain-additive-segment-contributions.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Callers declare an additive total and disjoint segments with the same unit, boundaries, and entity scope; incompatible or overlapping segments are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Missing or uncovered segments appear as an explicit remainder or unavailable decomposition; contribution shares are unavailable for a zero total change.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."},{"capabilityId":"investigation/cohorts/explain-additive-segment-contributions","title":"Find which segments contribute to an aggregate change","criterion":"Reports show observed segment changes against the declared comparison reference and reconcile their sum to the total change within stated tolerance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/cohorts/explain-additive-segment-contributions, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Locate contributors to a total change","order":20}},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","status":"planned","area":"relationships","path":".capabilities/relationships/cohorts/separate-rate-and-population-mix.capability.yaml","health":"planned","coverage":"uncovered","summary":"Decompose a declared aggregate rate change into within-cohort rate and cohort-weight changes using explicit aligned counts and a comparison period.","intent":"Explain aggregate conversion movements that can occur even when individual cohorts' performance is stable or improving.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"relationships/cohorts/separate-rate-and-population-mix has no automated or manual verification checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-automated-checks","message":"relationships/cohorts/separate-rate-and-population-mix has no automated checks","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"},{"code":"missing-manual-review","message":"relationships/cohorts/separate-rate-and-population-mix has no manual review guidance","capabilityId":"relationships/cohorts/separate-rate-and-population-mix","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\relationships\\cohorts\\separate-rate-and-population-mix.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Callers supply disjoint cohort numerators and denominators for explicit reference and comparison periods; invalid counts or overlap are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Empty, missing, new, or disappearing cohorts retain explicit support limits; unavailable components are not filled with zero or presented as causes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."},{"capabilityId":"relationships/cohorts/separate-rate-and-population-mix","title":"Distinguish changing performance from a changing population mix","criterion":"Rate and mix components reconcile to the aggregate change within stated tolerance; interaction allocation is visible rather than hidden.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to relationships/cohorts/separate-rate-and-population-mix, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Cohort investigation","backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","order":30}}],"deliveryStrategy":{"release":"Future - Cohort investigation","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/cohorts/analyze-declared-entity-batches"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/cohorts/analyze-declared-entity-batches has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Investigate changes across populations"],"missingBackbones":[],"stepCoverageGaps":["Investigate changes across populations > Locate contributors to a total change","Investigate changes across populations > Separate performance changes from population mix"],"rationale":"Starts with the earliest mapped step in Investigate changes across populations."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["investigation/cohorts/explain-additive-segment-contributions","relationships/cohorts/separate-rate-and-population-mix"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"investigation/cohorts/explain-additive-segment-contributions has verification gaps; relationships/cohorts/separate-rate-and-population-mix has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Investigate changes across populations"],"missingBackbones":[],"stepCoverageGaps":["Investigate changes across populations > Keep each entity's evidence separate"],"rationale":"Follows story-map order across Locate contributors to a total change, Separate performance changes from population mix."}]},"presentation":{"outcome":"Future - Cohort investigation helps teams move from Keep each entity's evidence separate to Separate performance changes from population mix across 1 backbone activity.","narrativePath":[{"backbone":"Investigate changes across populations","step":"Keep each entity's evidence separate","capabilityIds":["data/cohorts/analyze-declared-entity-batches"],"health":"planned"},{"backbone":"Investigate changes across populations","step":"Locate contributors to a total change","capabilityIds":["investigation/cohorts/explain-additive-segment-contributions"],"health":"planned"},{"backbone":"Investigate changes across populations","step":"Separate performance changes from population mix","capabilityIds":["relationships/cohorts/separate-rate-and-population-mix"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Continuous analysis","capabilities":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","status":"planned","area":"data","path":".capabilities/data/revisions/reconcile-late-and-corrected-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Accept explicit source revisions, recompute affected analysis and relationships, and retain the differences from previously published evidence.","intent":"Let callers correct history without silently overwriting what earlier analysis actually showed.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"data/revisions/reconcile-late-and-corrected-observations has no automated or manual verification checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-automated-checks","message":"data/revisions/reconcile-late-and-corrected-observations has no automated checks","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"},{"code":"missing-manual-review","message":"data/revisions/reconcile-late-and-corrected-observations has no manual review guidance","capabilityId":"data/revisions/reconcile-late-and-corrected-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\data\\revisions\\reconcile-late-and-corrected-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"A revision identifies the source, affected timestamps or positions, prior source version, replacement data, and declared completeness changes.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Invalid or conflicting revisions leave the previous state available; unchanged inputs preserve numerical evidence and unrelated source results.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."},{"capabilityId":"data/revisions/reconcile-late-and-corrected-observations","title":"See how late or corrected data changes the evidence","criterion":"Reanalysis includes downstream reference, calibration, pattern, and relationship effects rather than changing only the corrected sample.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to data/revisions/reconcile-late-and-corrected-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","order":30}},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","status":"planned","area":"execution","path":".capabilities/execution/incremental/append-finalized-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Append finalized observations to supported analysis state and return new evidence without reanalyzing unchanged history.","intent":"Reduce repeated work for caller-managed monitoring while preserving causal scoring and continuity of sustained-pattern evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/incremental/append-finalized-observations has no automated or manual verification checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-automated-checks","message":"execution/incremental/append-finalized-observations has no automated checks","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"},{"code":"missing-manual-review","message":"execution/incremental/append-finalized-observations has no manual review guidance","capabilityId":"execution/incremental/append-finalized-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\incremental\\append-finalized-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Callers append observations after the stored cutoff using unchanged source identity and settings; backdated changes require an explicit revision workflow.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Outputs identify newly available evidence, current cutoff, remaining support limitations, and resource-limit stops without silently skipping observations.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."},{"capabilityId":"execution/incremental/append-finalized-observations","title":"Update evidence when the next observations arrive","criterion":"Retrying an identified append batch does not duplicate samples or findings; failed appends leave the prior committed state available.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/incremental/append-finalized-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","order":20}},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","status":"planned","area":"execution","path":".capabilities/execution/state/save-and-restore-analysis-state.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export and restore versioned local analysis state with enough provenance to continue supported recipes without losing calibration or pattern history.","intent":"Let calling applications maintain continuity across process restarts while preserving reproducible evidence.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"execution/state/save-and-restore-analysis-state has no automated or manual verification checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-automated-checks","message":"execution/state/save-and-restore-analysis-state has no automated checks","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"},{"code":"missing-manual-review","message":"execution/state/save-and-restore-analysis-state has no manual review guidance","capabilityId":"execution/state/save-and-restore-analysis-state","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\execution\\state\\save-and-restore-analysis-state.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"A snapshot records its state format, engine version, supported recipe, settings, source identity, last finalized position, and numerical provenance.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Export requires an explicit caller destination and protects existing files; state retention and deletion remain caller-controlled.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."},{"capabilityId":"execution/state/save-and-restore-analysis-state","title":"Resume local analysis after a restart","criterion":"Restore checks integrity and compatibility and rejects corrupt, unsupported, or mismatched state with a structured reason.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to execution/state/save-and-restore-analysis-state, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Continuous analysis","backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","order":10}}],"deliveryStrategy":{"release":"Future - Continuous analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["execution/state/save-and-restore-analysis-state"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"execution/state/save-and-restore-analysis-state has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Keep evidence current as observations arrive"],"missingBackbones":[],"stepCoverageGaps":["Keep evidence current as observations arrive > Analyze the next finalized samples","Keep evidence current as observations arrive > Reconcile revisions without losing history"],"rationale":"Starts with the earliest mapped step in Keep evidence current as observations arrive."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["execution/incremental/append-finalized-observations","data/revisions/reconcile-late-and-corrected-observations"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"data/revisions/reconcile-late-and-corrected-observations has verification gaps; execution/incremental/append-finalized-observations has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Keep evidence current as observations arrive"],"missingBackbones":[],"stepCoverageGaps":["Keep evidence current as observations arrive > Preserve state across restarts"],"rationale":"Follows story-map order across Analyze the next finalized samples, Reconcile revisions without losing history."}]},"presentation":{"outcome":"Future - Continuous analysis helps teams move from Preserve state across restarts to Reconcile revisions without losing history across 1 backbone activity.","narrativePath":[{"backbone":"Keep evidence current as observations arrive","step":"Preserve state across restarts","capabilityIds":["execution/state/save-and-restore-analysis-state"],"health":"planned"},{"backbone":"Keep evidence current as observations arrive","step":"Analyze the next finalized samples","capabilityIds":["execution/incremental/append-finalized-observations"],"health":"planned"},{"backbone":"Keep evidence current as observations arrive","step":"Reconcile revisions without losing history","capabilityIds":["data/revisions/reconcile-late-and-corrected-observations"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Historical evaluation","capabilities":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","status":"planned","area":"evaluation","path":".capabilities/evaluation/comparisons/compare-monitor-configurations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Compare a bounded caller-supplied set of recipes, thresholds, and policies on the same historical replay and a separate evaluation interval.","intent":"Make the tradeoff between earlier detection and unnecessary investigation inspectable before a caller adopts new settings.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/comparisons/compare-monitor-configurations has no automated or manual verification checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/comparisons/compare-monitor-configurations has no automated checks","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/comparisons/compare-monitor-configurations has no manual review guidance","capabilityId":"evaluation/comparisons/compare-monitor-configurations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\comparisons\\compare-monitor-configurations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Callers declare candidate configurations, a selection interval, and a later non-overlapping evaluation interval; overlapping intervals are rejected.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Candidate reports use the same cutoffs, data, label coverage, and matching rules and expose failed or inapplicable candidates.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/comparisons/compare-monitor-configurations","title":"Choose settings using held-out historical evidence","criterion":"Comparisons report finding volume, reviewed-incident coverage, detection delay, and policy eligibility without merging overlapping evidence as independent wins.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/comparisons/compare-monitor-configurations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","order":30}},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","status":"planned","area":"evaluation","path":".capabilities/evaluation/outcomes/compare-findings-with-reviewed-incidents.capability.yaml","health":"planned","coverage":"uncovered","summary":"Match historical findings to caller-reviewed incident intervals and report detection timing and unmatched findings under explicit matching rules.","intent":"Help owners judge usefulness on representative history rather than treating synthetic examples or trigger counts as detection accuracy.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated or manual verification checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no automated checks","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/outcomes/compare-findings-with-reviewed-incidents has no manual review guidance","capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\outcomes\\compare-findings-with-reviewed-incidents.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Precision or recall is reported only for an explicitly reviewed evaluation population, with its counts and matching rules; unlabeled periods stay unknown.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reports distinguish matched incidents, missed labeled incidents, duplicate findings, unmatched findings, and periods without reviewed labels.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/outcomes/compare-findings-with-reviewed-incidents","title":"Measure detection timing against reviewed incidents","criterion":"Reviewed labels declare entity, incident interval, and matching tolerance; invalid intervals and ambiguous matches are surfaced for review.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/outcomes/compare-findings-with-reviewed-incidents, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","order":20}},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","status":"planned","area":"evaluation","path":".capabilities/evaluation/replay/replay-historical-observations.capability.yaml","health":"planned","coverage":"uncovered","summary":"Replay a supplied history through explicit observation cutoffs and retain the evidence available at each cutoff.","intent":"Establish when a change could first have been seen without using later observations to improve earlier results.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"evaluation/replay/replay-historical-observations has no automated or manual verification checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-automated-checks","message":"evaluation/replay/replay-historical-observations has no automated checks","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"},{"code":"missing-manual-review","message":"evaluation/replay/replay-historical-observations has no manual review guidance","capabilityId":"evaluation/replay/replay-historical-observations","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\evaluation\\replay\\replay-historical-observations.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Callers supply a history, fixed analysis settings, and ordered replay cutoffs; every replay result records its cutoff and configuration.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Each cutoff excludes later observations from season selection, training, calibration, relationship fitting, and evaluation.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."},{"capabilityId":"evaluation/replay/replay-historical-observations","title":"Replay history as it would have arrived","criterion":"Incomplete, missing, and insufficient-history states remain visible at each cutoff rather than being replaced by the final run's evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to evaluation/replay/replay-historical-observations, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Historical evaluation","backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","order":10}}],"deliveryStrategy":{"release":"Future - Historical evaluation","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["evaluation/replay/replay-historical-observations"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"evaluation/replay/replay-historical-observations has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Evaluate a monitor before relying on it"],"missingBackbones":[],"stepCoverageGaps":["Evaluate a monitor before relying on it > Compare detection with reviewed outcomes","Evaluate a monitor before relying on it > Review the tradeoff before adopting settings"],"rationale":"Starts with the earliest mapped step in Evaluate a monitor before relying on it."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["evaluation/outcomes/compare-findings-with-reviewed-incidents","evaluation/comparisons/compare-monitor-configurations"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"evaluation/comparisons/compare-monitor-configurations has verification gaps; evaluation/outcomes/compare-findings-with-reviewed-incidents has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Evaluate a monitor before relying on it"],"missingBackbones":[],"stepCoverageGaps":["Evaluate a monitor before relying on it > Reconstruct the evidence available at the time"],"rationale":"Follows story-map order across Compare detection with reviewed outcomes, Review the tradeoff before adopting settings."}]},"presentation":{"outcome":"Future - Historical evaluation helps teams move from Reconstruct the evidence available at the time to Review the tradeoff before adopting settings across 1 backbone activity.","narrativePath":[{"backbone":"Evaluate a monitor before relying on it","step":"Reconstruct the evidence available at the time","capabilityIds":["evaluation/replay/replay-historical-observations"],"health":"planned"},{"backbone":"Evaluate a monitor before relying on it","step":"Compare detection with reviewed outcomes","capabilityIds":["evaluation/outcomes/compare-findings-with-reviewed-incidents"],"health":"planned"},{"backbone":"Evaluate a monitor before relying on it","step":"Review the tradeoff before adopting settings","capabilityIds":["evaluation/comparisons/compare-monitor-configurations"],"health":"planned"}],"coverageSignals":[]}},{"release":"Future - Investigation memory","capabilities":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","status":"planned","area":"interfaces","path":".capabilities/interfaces/export/package-investigation-evidence.capability.yaml","health":"planned","coverage":"uncovered","summary":"Export a local portable investigation bundle containing selected evidence, analysis provenance, limitations, and supplied review records.","intent":"Make an investigation reviewable beyond a transient MCP session or an agent explanation without requiring a hosted service.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"interfaces/export/package-investigation-evidence has no automated or manual verification checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-automated-checks","message":"interfaces/export/package-investigation-evidence has no automated checks","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"},{"code":"missing-manual-review","message":"interfaces/export/package-investigation-evidence has no manual review guidance","capabilityId":"interfaces/export/package-investigation-evidence","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\interfaces\\export\\package-investigation-evidence.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"A readable local report links to included structured evidence and clearly distinguishes observations, statistical findings, and caller review judgments.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Bundles contain a manifest, settings and engine versions, source fingerprints, selected numerical evidence, readiness, execution status, and lineage.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."},{"capabilityId":"interfaces/export/package-investigation-evidence","title":"Share an investigation that can be inspected later","criterion":"Callers explicitly select runs or cases and an output destination; existing output requires explicit overwrite permission.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to interfaces/export/package-investigation-evidence, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","order":20}},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/record-reviewed-case-outcomes.capability.yaml","health":"planned","coverage":"uncovered","summary":"Attach caller-authored review decisions, context, and observed outcomes to versioned case evidence without rewriting numerical findings.","intent":"Build an inspectable record of which findings mattered and what reviewers learned for later replay and investigation.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/record-reviewed-case-outcomes has no automated or manual verification checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/record-reviewed-case-outcomes has no automated checks","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/record-reviewed-case-outcomes has no manual review guidance","capabilityId":"investigation/history/record-reviewed-case-outcomes","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\record-reviewed-case-outcomes.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"A review record identifies the case and evidence version, reviewer-supplied identity, review time, disposition, and supporting notes or references.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Corrections create an attributable review revision while retaining the earlier decision and original numerical evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/record-reviewed-case-outcomes","title":"Keep reviewed outcomes beside the original evidence","criterion":"Records distinguish confirmed incidents, expected changes, data issues, dismissed concerns, and unresolved cases without requiring a causal label.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/record-reviewed-case-outcomes, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","order":10}},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","status":"planned","area":"investigation","path":".capabilities/investigation/history/retrieve-comparable-reviewed-cases.capability.yaml","health":"planned","coverage":"uncovered","summary":"Search caller-selected local case records using explicit entity, metric, pattern, and review filters and show why returned cases are comparable.","intent":"Give investigators useful precedent while keeping similarity separate from a diagnosis of the current change.","references":{"total":0,"readable":0,"missing":[]},"verification":{"automated":0,"manual":0,"gaps":[{"code":"missing-verification","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated or manual verification checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-automated-checks","message":"investigation/history/retrieve-comparable-reviewed-cases has no automated checks","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"},{"code":"missing-manual-review","message":"investigation/history/retrieve-comparable-reviewed-cases has no manual review guidance","capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","filePath":"C:\\Users\\troym\\Git\\Anomalyizer\\.capabilities\\investigation\\history\\retrieve-comparable-reviewed-cases.capability.yaml"}]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":4,"assessor-limitation":0,"ignored":0},"nextAction":"Keep as roadmap context until implementation starts.","topFindings":[{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Missing context, incompatible units, and insufficient baseline support are visible and cannot be silently ranked as equivalent evidence.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Results expose the matched attributes, material differences, evidence versions, and review dispositions behind the comparison.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."},{"capabilityId":"investigation/history/retrieve-comparable-reviewed-cases","title":"Find earlier reviewed cases with comparable evidence","criterion":"Retrieval leaves the current result and policy unchanged and returns an explicit no-comparable-case result when the supplied population has no match.","evidence":[],"status":"no-implementation-reference","confidence":"high","action":"add-implementation-reference","rationale":"The capability has no agent.implementation.references, so the assessor has no concrete files to inspect.","recommendation":"Add agent.implementation.references to investigation/history/retrieve-comparable-reviewed-cases, or lower the status if this capability is only planned."}],"storyMap":{"release":"Future - Investigation memory","backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","order":30}}],"deliveryStrategy":{"release":"Future - Investigation memory","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["investigation/history/record-reviewed-case-outcomes"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"investigation/history/record-reviewed-case-outcomes has verification gaps","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Reuse what an investigation teaches us"],"missingBackbones":[],"stepCoverageGaps":["Reuse what an investigation teaches us > Inspect comparable reviewed history","Reuse what an investigation teaches us > Preserve a portable evidence record"],"rationale":"Starts with the earliest mapped step in Reuse what an investigation teaches us."},{"order":2,"phase":"mid-game","name":"Progressive capability layer","capabilityIds":["interfaces/export/package-investigation-evidence","investigation/history/retrieve-comparable-reviewed-cases"],"releaseStrategy":"Add the next coherent layer of user-visible capability after the walking skeleton is usable.","developmentStrategy":"Build in story-map order while preserving the release narrative and avoiding isolated deep work.","riskIntent":"interfaces/export/package-investigation-evidence has verification gaps; investigation/history/retrieve-comparable-reviewed-cases has verification gaps","learningIntent":"Learn which adjacent steps increase outcome value before investing in optimization or hardening.","backboneCoverage":["Reuse what an investigation teaches us"],"missingBackbones":[],"stepCoverageGaps":["Reuse what an investigation teaches us > Record the reviewed outcome"],"rationale":"Follows story-map order across Inspect comparable reviewed history, Preserve a portable evidence record."}]},"presentation":{"outcome":"Future - Investigation memory helps teams move from Record the reviewed outcome to Inspect comparable reviewed history across 1 backbone activity.","narrativePath":[{"backbone":"Reuse what an investigation teaches us","step":"Record the reviewed outcome","capabilityIds":["investigation/history/record-reviewed-case-outcomes"],"health":"planned"},{"backbone":"Reuse what an investigation teaches us","step":"Preserve a portable evidence record","capabilityIds":["interfaces/export/package-investigation-evidence"],"health":"planned"},{"backbone":"Reuse what an investigation teaches us","step":"Inspect comparable reviewed history","capabilityIds":["investigation/history/retrieve-comparable-reviewed-cases"],"health":"planned"}],"coverageSignals":[]}},{"release":"Incomplete-period monitoring","capabilities":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","status":"implemented","area":"analysis","path":".capabilities/analysis/resolution/monitor-within-unfinished-days.capability.yaml","health":"review","coverage":"partial","summary":"Multi-resolution analysis keeps native finalized subperiod evidence and completed fixed 24-hour aggregates, with a separate marker for the unfinished day.","intent":"Expose changes at the supplied sample cadence before a chart's daily total is complete.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":5,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A source cadence that does not evenly divide 24 hours is rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"A trailing partial aggregate reports observed progress and links its native evidence while remaining outside the completed-period series.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/resolution/monitor-within-unfinished-days","title":"See changes before the day is complete","criterion":"An explicitly unfinished source sample is excluded from both analysis views.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/resolution/monitor-within-unfinished-days --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","order":20}},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","status":"implemented","area":"data","path":".capabilities/data/quality/analyze-sparse-observations.capability.yaml","health":"review","coverage":"partial","summary":"Schema-1.1 callers declare coverage and missing-data semantics per source, keeping unknown collection distinct from confirmed empty additive volume.","intent":"Retain useful rule and supported statistical evidence for sparse histories while preserving time, provenance, incomplete periods and evidence limits.","references":{"total":10,"readable":10,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"As-of replay respects supplied arrival/revision timestamps and sparse multi-resolution views retain useful siblings without inventing complete totals.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":13,"excerpt":"\"seasonal-residual-v1\", \"adaptive-seasonal-v1\", \"multi-resolution-v1\","},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":304,"excerpt":"if (self.data_policy or self.coverage or self.available_at or self.revisions) and self.timestamps is None:"},{"reference":"src/anomalyzer/contracts.py","line":305,"excerpt":"raise ValueError(\"source policies/coverage/revisions require original timestamps\")"},{"reference":"src/anomalyzer/contracts.py","line":310,"excerpt":"if self.timestamps is None or self.data_policy is None or self.data_policy.observation_basis != \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":67,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse.py","line":69,"excerpt":"if coverage != \"complete\" or previous[\"coverage\"] != \"complete\" or ds.available_at or ds.revisions:"},{"reference":"src/anomalyzer/sparse.py","line":86,"excerpt":"observed_at = timestamp(revision.timestamp, ds.timezone)"},{"reference":"src/anomalyzer/sparse_views.py","line":1,"excerpt":"\"\"\"Schema-1.1 native and complete-day evidence with independent view eligibility.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":28,"excerpt":"case.revision_id = f\"{resolution}:{case.revision_id}\""},{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/orchestrator.py","line":107,"excerpt":"if request.config.recipe == \"multi-resolution-v1\":"},{"reference":"src/anomalyzer/orchestrator.py","line":108,"excerpt":"from .sparse_views import analyze_multi_request"},{"reference":"src/anomalyzer/orchestrator.py","line":134,"excerpt":"if \"data_policy\" in source.quality and request.config.recipe != \"multi-resolution-v1\""},{"reference":"src/anomalyzer/orchestrator.py","line":170,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":195,"excerpt":"active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of)),"},{"reference":"src/anomalyzer/relationships.py","line":164,"excerpt":"source_timestamps = {source.dataset.id: source.times[source_index]"},{"reference":"src/anomalyzer/relationships.py","line":166,"excerpt":"cutoff = max(source_timestamps.values())"},{"reference":"src/anomalyzer/relationships.py","line":170,"excerpt":"source_indexes=source_indexes, source_timestamps=source_timestamps,"},{"reference":"src/anomalyzer/relationships.py","line":186,"excerpt":"lineage[-1].source_original_timestamps = {source.dataset.id: source.quality[\"samples\"][i][\"original_timestamp\"] for source, i in used}"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":124,"excerpt":"def test_sparse_phase_without_support_is_explicitly_unavailable():"},{"reference":"tests/test_sparse.py","line":181,"excerpt":"ds[\"timestamps\"][1] = \"2026-02-01T00:00:00Z\""},{"reference":"docs/sparse-data.md","line":31,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":74,"excerpt":"conversion reasons, counts, revision IDs and availability timestamps. Converted"},{"reference":"docs/sparse-data.md","line":88,"excerpt":"otherwise bounds come from the supplied as-of-available timestamp range."},{"reference":"docs/sparse-data.md","line":106,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"examples/sparse_examples.py","line":59,"excerpt":"daily = request(dataset([1.0]*60, data_policy={\"measurement\":\"additive_volume\"}), recipe=\"multi-resolution-v1\","},{"reference":"examples/sparse_examples.py","line":67,"excerpt":"\"timestamps\":[\"2026-01-01T09:00:00.123Z\"]*3+[\"2026-01-01T09:00:05.456Z\"]*4+[\"2026-01-01T09:00:10.789Z\"]*3,"},{"reference":"examples/sparse_examples.py","line":74,"excerpt":"corrected[\"datasets\"][0][\"revisions\"] = [{\"event_id\":\"call-07\", \"timestamp\":same_time[\"timestamps\"][7],"},{"reference":"examples/sparse_examples.py","line":83,"excerpt":"timestamps=same_time[\"timestamps\"][:7]+[same_time[\"timestamps\"][-1]]*12,"},{"reference":"examples/sparse_examples.py","line":96,"excerpt":"for name in (\"timestamps\", \"values\", \"event_ids\"):"},{"reference":"examples/render_readme_progression.py","line":686,"excerpt":"def figure_multi_resolution():"},{"reference":"examples/render_readme_progression.py","line":699,"excerpt":"\"config\": {\"recipe\": \"multi-resolution-v1\", \"aggregate_function\": \"sum\"},"},{"reference":"examples/render_readme_progression.py","line":710,"excerpt":"assert partial[\"included_in_completed_period_view\"] is False"},{"reference":"examples/render_readme_progression.py","line":767,"excerpt":"complete_count = result.data_quality[\"views\"][\"completed_period\"][\"observation_count\"]"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Bounded calendar preparation preserves phase, original values and coverage; only an explicit additive-volume empty policy authorizes zero conversion.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":51,"excerpt":"def valid_reset_points(cls, values):"},{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":71,"excerpt":"def numeric_values(cls, data):"},{"reference":"src/anomalyzer/contracts.py","line":85,"excerpt":"if self.period_statuses is not None and len(self.period_statuses) != len(self.values):"},{"reference":"src/anomalyzer/contracts.py","line":86,"excerpt":"raise ValueError(\"period_statuses and values must have equal length\")"},{"reference":"src/anomalyzer/sparse.py","line":3,"excerpt":"Calendar coordinates remain dense only inside an explicitly bounded input grid."},{"reference":"src/anomalyzer/sparse.py","line":34,"excerpt":"if ds.data_policy or ds.coverage is not None or ds.available_at is not None or ds.revisions or any(v is None for v in ds.values) or \"incomplete\" in (ds.period_statuses or []):"},{"reference":"src/anomalyzer/sparse.py","line":42,"excerpt":"\"\"\"Return the explicit grid, eligible model values and inspectable provenance.\"\"\""},{"reference":"src/anomalyzer/sparse.py","line":51,"excerpt":"statuses = ds.period_statuses or [\"complete\"] * len(ds.values)"},{"reference":"src/anomalyzer/sparse.py","line":57,"excerpt":"\"complete\" if value is not None else \"empty\" if policy.absence_means_zero else \"unknown\")"},{"reference":"src/anomalyzer/sparse_views.py","line":72,"excerpt":"measurements = quality[\"rule_values\"]"},{"reference":"src/anomalyzer/sparse_views.py","line":75,"excerpt":"if policy.observation_basis != \"calendar\":"},{"reference":"src/anomalyzer/sparse_views.py","line":82,"excerpt":"if policy.measurement != \"additive_volume\" or cfg.aggregate_function != \"sum\":"},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":98,"excerpt":"daily_times, daily_values, coverage, records, available_at = [], [], [], [], []"},{"reference":"src/anomalyzer/orchestrator.py","line":29,"excerpt":"if not any(row.get(\"available_at\") and (row.get(\"value\") is not None or row.get(\"coverage\") == \"excluded\") for row in samples):"},{"reference":"src/anomalyzer/orchestrator.py","line":64,"excerpt":"unavailable values. Dropping only that leading run lets the ordinary engine"},{"reference":"src/anomalyzer/orchestrator.py","line":128,"excerpt":"prepared_points += len(source.values) + len(dataset.revisions)"},{"reference":"src/anomalyzer/orchestrator.py","line":140,"excerpt":"source.quality.get(\"rule_values\", source.values)[:source.quality.get(\"grid_count\", source.quality[\"observation_count\"])], dataset.rules,"},{"reference":"src/anomalyzer/orchestrator.py","line":206,"excerpt":"[dict(value=p.value, available_at=p.latest_source_cutoff if any(p.source_available_at.values()) else None) for p in lineage])"},{"reference":"src/anomalyzer/relationships.py","line":40,"excerpt":"sources = [replace(source, values=source.quality.get(\"rule_values\", source.values)) for source in sources]"},{"reference":"src/anomalyzer/relationships.py","line":62,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":102,"excerpt":"value, unavailable = None, \"zero_denominator\""},{"reference":"src/anomalyzer/relationships.py","line":135,"excerpt":"y, x = response.values[index], predictor.values[predictor_index]"},{"reference":"src/anomalyzer/relationships.py","line":145,"excerpt":"source_values = [source.values[source_index]"},{"reference":"src/anomalyzer/early.py","line":58,"excerpt":"for index, (observed_at, value) in enumerate(zip(times, values)):"},{"reference":"src/anomalyzer/early.py","line":137,"excerpt":"criterion_met and policy.allow_explicit_rules)"},{"reference":"tests/test_sparse.py","line":45,"excerpt":"def test_zero_is_only_authorized_for_complete_empty_additive_volume():"},{"reference":"tests/test_sparse.py","line":46,"excerpt":"request = body([2, None, None, 0, 9], data_policy={\"missing\": \"zero_for_empty\", \"measurement\": \"additive_volume\"},"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":52,"excerpt":"assert rows[1][\"reason\"] == \"confirmed_empty_additive_volume\""},{"reference":"tests/test_sparse.py","line":58,"excerpt":"def test_global_collection_contract_authorizes_omitted_slots_but_not_explicit_unknown():"},{"reference":"docs/sparse-data.md","line":4,"excerpt":"Schema 1.1 therefore preserves missing values by default. An expectation can be"},{"reference":"docs/sparse-data.md","line":34,"excerpt":"data_policy=DataPolicy(),  # preserve missing; hour 2 is also unknown"},{"reference":"docs/sparse-data.md","line":42,"excerpt":"| Numeric value, including zero, `coverage=\"complete\"` | A measurement eligible for rules and model support. Explicit zero remains zero. |"},{"reference":"docs/sparse-data.md","line":44,"excerpt":"| `coverage=\"empty\"` | Caller confirms a finalized period with no reported volume. Preserve as missing unless an additive zero policy explicitly authorizes conversion. |"},{"reference":"docs/sparse-data.md","line":45,"excerpt":"| `coverage=\"unknown\"` | Collection/exposure is unknown; never filled even under a global empty-volume contract. |"},{"reference":"examples/sparse_examples.py","line":22,"excerpt":"preserve = request(dataset(values, coverage=[\"empty\" if v is None else \"complete\" for v in values],"},{"reference":"examples/sparse_examples.py","line":23,"excerpt":"data_policy={\"measurement\": \"additive_volume\"}))"},{"reference":"examples/sparse_examples.py","line":25,"excerpt":"zero[\"datasets\"][0][\"data_policy\"][\"missing\"] = \"zero_for_empty\""},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":28,"excerpt":"coverage=[\"complete\"]*2+[\"unknown\"]+[\"complete\"]*6+[\"unknown\"]+[\"complete\"]*3+[\"incomplete\"], data_policy={}))"},{"reference":"examples/render_readme_progression.py","line":135,"excerpt":"def chart(svg, values, overlays, x, y, width, height, *, y_bounds=None,"},{"reference":"examples/render_readme_progression.py","line":139,"excerpt":"all_values = [values] + [series for _, series, *_ in overlays]"},{"reference":"examples/render_readme_progression.py","line":140,"excerpt":"low, high = y_bounds or bounds(*all_values)"},{"reference":"examples/render_readme_progression.py","line":166,"excerpt":"if value is not None and (index == 0 or values[index-1] is None) and (index == count-1 or values[index+1] is None):"},{"reference":"examples/render_readme_progression.py","line":206,"excerpt":"values = [80 + 0.65 * index + [0, 0.25, -0.15, 0.10, -0.20][index % 5]"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"data/quality/analyze-sparse-observations","title":"Assess sparse observations without inventing consumption","criterion":"Event-index profiles preserve real event timestamps and reject calendar seasonality; unavailable relationship inputs and zero denominators stay unavailable.","evidence":[{"reference":"src/anomalyzer/contracts.py","line":54,"excerpt":"raise ValueError(\"reset_points must contain nonempty timestamps or integer sample positions\")"},{"reference":"src/anomalyzer/contracts.py","line":81,"excerpt":"elif self.frequency is None and getattr(getattr(self, \"data_policy\", None), \"observation_basis\", None) != \"event_index\":"},{"reference":"src/anomalyzer/contracts.py","line":242,"excerpt":"missing: Literal[\"preserve\", \"zero_for_empty\"] = \"preserve\""},{"reference":"src/anomalyzer/contracts.py","line":245,"excerpt":"observation_basis: Literal[\"calendar\", \"event_index\"] = \"calendar\""},{"reference":"src/anomalyzer/contracts.py","line":246,"excerpt":"event_order: Literal[\"timestamp\", \"trusted_sequence\"] = \"timestamp\""},{"reference":"src/anomalyzer/sparse.py","line":4,"excerpt":"Event coordinates retain actual timestamps and represent comparable events, not time."},{"reference":"src/anomalyzer/sparse.py","line":52,"excerpt":"for index, (text, value, status) in enumerate(zip(ds.timestamps, ds.values, statuses)):"},{"reference":"src/anomalyzer/sparse.py","line":60,"excerpt":"available = timestamp(ds.available_at[index], ds.timezone) if ds.available_at and ds.available_at[index] else None"},{"reference":"src/anomalyzer/sparse.py","line":66,"excerpt":"if cfg.duplicate_policy == \"error\" or policy.observation_basis == \"event_index\":"},{"reference":"src/anomalyzer/sparse.py","line":67,"excerpt":"raise ValueError(\"duplicate timestamps require declared duplicate_policy; event_index needs unique event timestamps\")"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":128,"excerpt":"policy_zero_children=sum(quality[\"samples\"][i][\"reason\"] == \"confirmed_empty_additive_volume\" for i in indexes),"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":134,"excerpt":"timestamps=daily_times, values=daily_values, frequency=\"1d\", coverage=coverage, available_at=available_at,"},{"reference":"src/anomalyzer/orchestrator.py","line":38,"excerpt":"assessment.baseline[\"known_input_availability_cutoff\"] = cutoffs[assessment.index]"},{"reference":"src/anomalyzer/orchestrator.py","line":39,"excerpt":"assessment.assumptions.append(\"Availability cutoff is conservative across known prior inputs; unknown arrival history cannot establish delivery-time detection performance.\")"},{"reference":"src/anomalyzer/orchestrator.py","line":170,"excerpt":"if any(definitions[name].timestamps is None for name in _source_ids(relationship)):"},{"reference":"src/anomalyzer/orchestrator.py","line":191,"excerpt":"relationship.id, derived.timestamps or [], derived.values,"},{"reference":"src/anomalyzer/orchestrator.py","line":195,"excerpt":"active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of)),"},{"reference":"src/anomalyzer/relationships.py","line":44,"excerpt":"raise ValueError(\"relationship timestamps cannot be positional\")"},{"reference":"src/anomalyzer/relationships.py","line":61,"excerpt":"for index in range(relationship.lag, coefficient_cutoff + 1):"},{"reference":"src/anomalyzer/relationships.py","line":62,"excerpt":"x = predictor.values[index - relationship.lag]"},{"reference":"src/anomalyzer/relationships.py","line":100,"excerpt":"if relationship.zero_denominator == \"error\":"},{"reference":"src/anomalyzer/relationships.py","line":101,"excerpt":"raise ValueError(f\"zero denominator at aligned index {index}\")"},{"reference":"src/anomalyzer/early.py","line":140,"excerpt":"unavailable_rules = [dict(id=rule.id, reason=\"no_comparable_prior\" if previous is None else \"relative_change_from_zero\")"},{"reference":"src/anomalyzer/early.py","line":147,"excerpt":"target_id=target_id, index=index, timestamp=observed_at,"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":68,"excerpt":"{\"observation_basis\": \"event_index\", \"grid_start\": \"2026-01-01T00:00:00Z\"},"},{"reference":"tests/test_sparse.py","line":76,"excerpt":"(lambda d: d[\"timestamps\"].__setitem__(2, \"2026-01-01T02:30:00Z\"), \"off\"),"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":129,"excerpt":"assert any(row[\"reason\"] == \"no_training_support_for_calendar_phase\" for row in result.methods[0].diagnostics[\"unavailable\"])"},{"reference":"docs/sparse-data.md","line":31,"excerpt":"timestamps=[\"2026-01-01T00:00:00Z\", \"2026-01-01T01:00:00Z\","},{"reference":"docs/sparse-data.md","line":106,"excerpt":"Duplicate timestamps normally fail. Legacy schema-1.1 relationship preparation"},{"reference":"docs/sparse-data.md","line":111,"excerpt":"rejected. Event-index timestamps without `event_ids` must be unique; identified"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":125,"excerpt":"timestamps=[\"2026-01-01T09:00:10.789Z\", \"2026-01-01T09:00:10.789Z\"],"},{"reference":"examples/sparse_examples.py","line":26,"excerpt":"result.update(preserve=preserve, zero_empty=zero)"},{"reference":"examples/sparse_examples.py","line":36,"excerpt":"events = dataset(costs, units=\"dollars_per_call\", data_policy={\"observation_basis\": \"event_index\"})"},{"reference":"examples/sparse_examples.py","line":38,"excerpt":"events[\"timestamps\"] = [(datetime(2026,1,1,tzinfo=timezone.utc)+timedelta(hours=h)).isoformat() for h in hours]"},{"reference":"examples/sparse_examples.py","line":57,"excerpt":"relation[\"relationships\"] = [{\"id\":\"unit-cost\",\"kind\":\"ratio\",\"numerator\":\"series\",\"denominator\":\"calls\",\"units\":\"dollars_per_call\"}]"},{"reference":"examples/sparse_examples.py","line":67,"excerpt":"\"timestamps\":[\"2026-01-01T09:00:00.123Z\"]*3+[\"2026-01-01T09:00:05.456Z\"]*4+[\"2026-01-01T09:00:10.789Z\"]*3,"},{"reference":"examples/render_readme_progression.py","line":654,"excerpt":"svg.text(430, 467, \"Sample index (zero-based)\", 15, MUTED, anchor=\"middle\")"},{"reference":"examples/render_readme_progression.py","line":697,"excerpt":"\"timestamps\": [(origin + timedelta(hours=index)).isoformat()"},{"reference":"examples/render_readme_progression.py","line":884,"excerpt":"header(svg, \"14\", \"Choose what an empty period means\", \"Identical source timestamps and values; only the declared empty-data policy changes.\")"},{"reference":"examples/render_readme_progression.py","line":887,"excerpt":"shared_high = max(v for v in observed(\"preserve\")+expectation(\"preserve\")+observed(\"zero_empty\")+expectation(\"zero_empty\") if v is not None)*1.12"},{"reference":"examples/render_readme_progression.py","line":897,"excerpt":"header(svg, \"15\", \"Keep zero, unknown and unfinished distinct\", \"Blue zeros are actual measurements; gaps stay gaps; the orange open-period value is excluded.\")"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Assess sparse source data","order":30}},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","status":"implemented","area":"data","path":".capabilities/data/quality/exclude-incomplete-periods.capability.yaml","health":"review","coverage":"partial","summary":"Explicitly incomplete observations are reported as data quality evidence and excluded from completed-series analysis.","intent":"Prevent an unfinished period from appearing to be an unexplained drop.","references":{"total":8,"readable":8,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"A relationship with an unavailable latest source reports that unavailability rather than normal behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Incomplete source samples are excluded from expectations, calibration, point scoring, and completed aggregates.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"data/quality/exclude-incomplete-periods","title":"Avoid mistaking unfinished periods for drops","criterion":"Schema-1.0 incomplete observations form a trailing suffix; schema-1.1 source policies preserve internal and trailing incomplete positions as unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify data/quality/exclude-incomplete-periods --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Incomplete-period monitoring","backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","order":10}}],"deliveryStrategy":{"release":"Incomplete-period monitoring","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["data/quality/exclude-incomplete-periods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"data/quality/exclude-incomplete-periods needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Monitor while periods are still open"],"missingBackbones":[],"stepCoverageGaps":["Monitor while periods are still open > Assess sparse source data","Monitor while periods are still open > Compare finalized subperiods"],"rationale":"Starts with the earliest mapped step in Monitor while periods are still open."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/resolution/monitor-within-unfinished-days","data/quality/analyze-sparse-observations"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/resolution/monitor-within-unfinished-days needs review; data/quality/analyze-sparse-observations needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Monitor while periods are still open"],"missingBackbones":[],"stepCoverageGaps":["Monitor while periods are still open > Keep unfinished data visible"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/resolution/monitor-within-unfinished-days, data/quality/analyze-sparse-observations."}]},"presentation":{"outcome":"Incomplete-period monitoring helps teams move from Keep unfinished data visible to Assess sparse source data across 1 backbone activity.","narrativePath":[{"backbone":"Monitor while periods are still open","step":"Keep unfinished data visible","capabilityIds":["data/quality/exclude-incomplete-periods"],"health":"review"},{"backbone":"Monitor while periods are still open","step":"Compare finalized subperiods","capabilityIds":["analysis/resolution/monitor-within-unfinished-days"],"health":"review"},{"backbone":"Monitor while periods are still open","step":"Assess sparse source data","capabilityIds":["data/quality/analyze-sparse-observations"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Monitor while periods are still open > Assess sparse source data","message":"Monitor while periods are still open > Assess sparse source data has needs-review coverage."},{"kind":"weak","label":"Monitor while periods are still open > Compare finalized subperiods","message":"Monitor while periods are still open > Compare finalized subperiods has needs-review coverage."},{"kind":"weak","label":"Monitor while periods are still open > Keep unfinished data visible","message":"Monitor while periods are still open > Keep unfinished data visible has needs-review coverage."}]}},{"release":"Python integration","capabilities":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/analyze-through-public-api.capability.yaml","health":"review","coverage":"partial","summary":"The public Python API accepts series or canonical requests and returns the same numerical analysis used by the CLI and MCP compatibility interfaces.","intent":"Embed local analysis in an application without changing the meaning of the evidence.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze accepts bare ordered values with separate validated settings and dispatches canonical schema 1.0 or 1.1 requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_relationships preserves the declared multi-dataset relationship semantics.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/python/analyze-through-public-api","title":"Embed local analysis in my application","criterion":"analyze_series preserves supplied settings and event-time context for a single dataset.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/python/analyze-through-public-api --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Call the public API","order":10}},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","status":"implemented","area":"interfaces","path":".capabilities/interfaces/python/install-versioned-library.capability.yaml","health":"review","coverage":"partial","summary":"Applications can install an identified Anomalyzer distribution and run the public analysis API locally with bundled dependencies.","intent":"Replace copied source files with a reviewable dependency that can be pinned and rolled back independently of application deployments.","references":{"total":11,"readable":11,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"A Linux Python 3.12 build bundles compatible binary dependencies and executes the consumer example with networking disabled.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/publish_release.py","line":54,"excerpt":"\"Linux/Python 3.12 Lambda consumer verified offline. \""},{"reference":"scripts/verify_distribution.py","line":47,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":6,"excerpt":"bundle=\"$root/.tmp/lambda-consumer\""},{"reference":"scripts/verify_lambda.sh","line":18,"excerpt":"python -m pip download --only-binary=:all: --require-hashes \\"},{"reference":"scripts/verify_lambda.sh","line":20,"excerpt":"python -m pip install --no-index --only-binary=:all: --require-hashes \\"},{"reference":".github/workflows/ci.yml","line":29,"excerpt":"- run: uv run python examples/generate.py"},{"reference":".github/workflows/ci.yml","line":32,"excerpt":"- run: uv run python examples/run_demo.py"},{"reference":".github/workflows/ci.yml","line":33,"excerpt":"- run: uv run python examples/run_agentic_demo.py"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/ci.yml","line":54,"excerpt":"run: uv export --locked --no-dev --no-emit-project --output-file dist/runtime-dependencies.txt"},{"reference":".github/workflows/release.yml","line":62,"excerpt":"- name: Verify the downloaded release in the offline Lambda-compatible bundle"},{"reference":"docs/releases/README.md","line":16,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":62,"excerpt":"uv run python examples/run_demo.py"},{"reference":"docs/releases/README.md","line":63,"excerpt":"uv run python examples/run_agentic_demo.py"},{"reference":"docs/releases/README.md","line":79,"excerpt":"versions, and tests a bundled Linux x86_64/Python 3.12 consumer in AWS's Lambda"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Clean consumers outside the source checkout can install either distribution and exercise contracts, recipes, assessments, relationships, readiness and cases.","evidence":[{"reference":"scripts/verify_distribution.py","line":1,"excerpt":"\"\"\"Install wheel AND sdist in clean environments outside the source tree (requires uv).\"\"\""},{"reference":"scripts/verify_distribution.py","line":48,"excerpt":"run([\"uv\", \"pip\", \"install\", \"--python\", python, \"--no-deps\", artifact], outside, env=env)"},{"reference":"examples/library/guard_smoke.py","line":32,"excerpt":"analyze_relationships, get_case, replay_policy,"},{"reference":"examples/library/guard_smoke.py","line":49,"excerpt":"assert legacy.methods[0].diagnostics[\"detection_readiness\"]"},{"reference":"examples/library/guard_smoke.py","line":65,"excerpt":"\"relationships\": [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"cost\","},{"reference":"examples/library/guard_smoke.py","line":69,"excerpt":"result = analyze_relationships(request)"},{"reference":"examples/library/guard_smoke.py","line":76,"excerpt":"short = analyze_relationships({**request.model_dump(mode=\"json\"),"},{"reference":".github/workflows/ci.yml","line":59,"excerpt":"- name: Test built wheel and sdist outside checkout"},{"reference":".github/workflows/release.yml","line":60,"excerpt":"- name: Install the exact published wheel URL with hashes outside checkout"},{"reference":"docs/releases/README.md","line":17,"excerpt":"| Source distribution (`.tar.gz`, or sdist) | Source needed to build the package. Useful for rebuilding and inspection; installation also needs the declared build backend. |"},{"reference":"docs/releases/README.md","line":99,"excerpt":"assets, installs the exact public wheel URL with its hashes in a clean consumer,"},{"reference":"docs/releases/README.md","line":217,"excerpt":"`analyze_series`, `analyze_relationships`, `get_case`, `replay_policy` (also"},{"reference":"docs/releases/README.md","line":220,"excerpt":"`anomalyzer.contracts`. See the self-contained"},{"reference":"docs/releases/README.md","line":222,"excerpt":"all APIs, recipes, point/pattern evidence, relationships, cases and policy"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"interfaces/python/install-versioned-library","title":"Install a versioned analysis library","criterion":"Version tags agree with package metadata and published release assets include SHA-256 checksums, source revision and pinned transitive dependencies.","evidence":[{"reference":"pyproject.toml","line":26,"excerpt":"dependencies = [\"pydantic>=2.10,<3\", \"typing-extensions>=4.12.2\", \"tzdata>=2024.1; sys_platform == 'win32'\"]"},{"reference":"pyproject.toml","line":33,"excerpt":"[project.optional-dependencies]"},{"reference":"scripts/release.py","line":34,"excerpt":"return hashlib.sha256(path.read_bytes()).hexdigest()"},{"reference":"scripts/release.py","line":81,"excerpt":"runtime = dist / \"runtime-dependencies.txt\""},{"reference":"scripts/release.py","line":82,"excerpt":"assert runtime.is_file(), \"export locked runtime dependencies first\""},{"reference":"scripts/release.py","line":85,"excerpt":"suffix = f\" --hash=sha256:{digest(wheel)}\\n\""},{"reference":"scripts/release.py","line":92,"excerpt":"assets = sorted(p for p in dist.iterdir() if p.is_file()"},{"reference":"scripts/publish_release.py","line":1,"excerpt":"\"\"\"Publish tested assets without replacing tags or existing assets (GitHub CLI auth).\"\"\""},{"reference":"scripts/publish_release.py","line":18,"excerpt":"def find_release(tag):"},{"reference":"scripts/publish_release.py","line":19,"excerpt":"\"\"\"The by-tag endpoint omits drafts; the authenticated release list includes them.\"\"\""},{"reference":"scripts/publish_release.py","line":20,"excerpt":"endpoint = f\"repos/{REPOSITORY}/releases/tags/{tag}\""},{"reference":"scripts/publish_release.py","line":40,"excerpt":"remote_sha = json.loads(gh(\"api\", f\"repos/{REPOSITORY}/commits/{tag}\"))[\"sha\"]"},{"reference":"scripts/verify_distribution.py","line":47,"excerpt":"dist / \"runtime-dependencies.txt\"], outside, env=env)"},{"reference":"scripts/verify_lambda.sh","line":25,"excerpt":"sha256sum *.whl > /output/wheelhouse-SHA256SUMS"},{"reference":"examples/library/guard_smoke.py","line":39,"excerpt":"assert anomalyzer.__version__ == importlib.metadata.version(\"anomalyzer\")"},{"reference":".github/workflows/ci.yml","line":44,"excerpt":"- name: Check release tag before building"},{"reference":".github/workflows/ci.yml","line":47,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/ci.yml","line":48,"excerpt":"run: python scripts/release.py check-version --tag \"$RELEASE_TAG\""},{"reference":".github/workflows/ci.yml","line":51,"excerpt":"- name: Check package metadata"},{"reference":".github/workflows/ci.yml","line":53,"excerpt":"- name: Export locked runtime dependencies"},{"reference":".github/workflows/release.yml","line":34,"excerpt":"name: release-dist-${{ github.sha }}"},{"reference":".github/workflows/release.yml","line":36,"excerpt":"- name: Stage tested assets for owner publication"},{"reference":".github/workflows/release.yml","line":39,"excerpt":"RELEASE_TAG: ${{ github.ref_name }}"},{"reference":".github/workflows/release.yml","line":40,"excerpt":"run: python scripts/publish_release.py --tag \"$RELEASE_TAG\" --revision \"$GITHUB_SHA\""},{"reference":".github/workflows/release.yml","line":48,"excerpt":"ref: ${{ github.event.release.tag_name }}"},{"reference":"docs/releases/README.md","line":8,"excerpt":"The original `v0.2.0` tag remains an unpublished candidate after a draft-lookup"},{"reference":"docs/releases/README.md","line":16,"excerpt":"| Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |"},{"reference":"docs/releases/README.md","line":18,"excerpt":"| Version tag (`v0.3.0`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |"},{"reference":"docs/releases/README.md","line":19,"excerpt":"| Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |"},{"reference":"docs/releases/README.md","line":23,"excerpt":"it for package metadata and the public `__version__` imports it."},{"reference":"tests/test_release.py","line":1,"excerpt":"\"\"\"Release safety: reject wrong tags and refuse changes to published artifacts.\"\"\""},{"reference":"tests/test_release.py","line":21,"excerpt":"def test_release_refuses_mismatched_or_unsupported_tag(tag):"},{"reference":"tests/test_release.py","line":23,"excerpt":"release.check_tag(tag)"},{"reference":"tests/test_release.py","line":52,"excerpt":"def test_published_matching_release_is_noop(monkeypatch, tmp_path):"},{"reference":"tests/test_release.py","line":81,"excerpt":"\"version\": release.version(), \"source_revision\": \"a\" * 40, \"tag\": f\"v{release.version()}\","}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Python integration","backbone":"Embed analysis in my application","step":"Install a released dependency","order":20}}],"deliveryStrategy":{"release":"Python integration","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/python/analyze-through-public-api"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/python/analyze-through-public-api needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Embed analysis in my application"],"missingBackbones":[],"stepCoverageGaps":["Embed analysis in my application > Install a released dependency"],"rationale":"Starts with the earliest mapped step in Embed analysis in my application."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/python/install-versioned-library"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/python/install-versioned-library needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Embed analysis in my application"],"missingBackbones":[],"stepCoverageGaps":["Embed analysis in my application > Call the public API"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/python/install-versioned-library."}]},"presentation":{"outcome":"Python integration helps teams move from Call the public API to Install a released dependency across 1 backbone activity.","narrativePath":[{"backbone":"Embed analysis in my application","step":"Call the public API","capabilityIds":["interfaces/python/analyze-through-public-api"],"health":"review"},{"backbone":"Embed analysis in my application","step":"Install a released dependency","capabilityIds":["interfaces/python/install-versioned-library"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Embed analysis in my application > Call the public API","message":"Embed analysis in my application > Call the public API has needs-review coverage."},{"kind":"weak","label":"Embed analysis in my application > Install a released dependency","message":"Embed analysis in my application > Install a released dependency has needs-review coverage."}]}},{"release":"Read-only MCP server","capabilities":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/connect-read-only-local-server.capability.yaml","health":"review","coverage":"partial","summary":"The local stdio MCP server advertises three read-only investigation tools with schemas and interpretation guidance.","intent":"Allow agent clients to discover and call the analysis tools using a clear local protocol contract.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Analysis tool calls return structured content together with compact text.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Invalid tool input and unknown session run IDs return structured errors rather than fabricated results.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/connect-read-only-local-server","title":"Give an agent read-only access to local analysis","criterion":"Legacy initialization and modern stateless discovery return supported protocol metadata and interpretation guidance.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/connect-read-only-local-server --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Connect a local server","order":10}},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","status":"implemented","area":"interfaces","path":".capabilities/interfaces/mcp/investigate-findings-and-evidence.capability.yaml","health":"review","coverage":"partial","summary":"Agents can analyze a request, page through its findings, and retrieve supporting samples, assessments, readiness, and relationship lineage.","intent":"Keep the initial agent response compact while making the supporting evidence inspectable.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"get_evidence exposes the selected finding's numerical support and relationship source lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"list_findings pages through cases, candidates, episodes, patterns, and data-quality findings.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"interfaces/mcp/investigate-findings-and-evidence","title":"Let an agent inspect the evidence behind a finding","criterion":"The analyze tool returns a compact headline, finding preview, and run ID for subsequent evidence requests.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify interfaces/mcp/investigate-findings-and-evidence --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Read-only MCP server","backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","order":20}}],"deliveryStrategy":{"release":"Read-only MCP server","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["interfaces/mcp/connect-read-only-local-server"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"interfaces/mcp/connect-read-only-local-server needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Give an agent access to local evidence"],"missingBackbones":[],"stepCoverageGaps":["Give an agent access to local evidence > Drill into findings and evidence"],"rationale":"Starts with the earliest mapped step in Give an agent access to local evidence."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["interfaces/mcp/investigate-findings-and-evidence"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"interfaces/mcp/investigate-findings-and-evidence needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Give an agent access to local evidence"],"missingBackbones":[],"stepCoverageGaps":["Give an agent access to local evidence > Connect a local server"],"rationale":"Prioritizes capabilities with delivery risk signals: interfaces/mcp/investigate-findings-and-evidence."}]},"presentation":{"outcome":"Read-only MCP server helps teams move from Connect a local server to Drill into findings and evidence across 1 backbone activity.","narrativePath":[{"backbone":"Give an agent access to local evidence","step":"Connect a local server","capabilityIds":["interfaces/mcp/connect-read-only-local-server"],"health":"review"},{"backbone":"Give an agent access to local evidence","step":"Drill into findings and evidence","capabilityIds":["interfaces/mcp/investigate-findings-and-evidence"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Give an agent access to local evidence > Connect a local server","message":"Give an agent access to local evidence > Connect a local server has needs-review coverage."},{"kind":"weak","label":"Give an agent access to local evidence > Drill into findings and evidence","message":"Give an agent access to local evidence > Drill into findings and evidence has needs-review coverage."}]}},{"release":"Related-metric analysis","capabilities":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/compare-declared-metrics.capability.yaml","health":"review","coverage":"partial","summary":"Explicit relationships produce ratio, difference, normalized-residual, and joint-condition series with source-level lineage.","intent":"Reveal changes between metrics that individual source values may not expose.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Derived series can be analyzed with the frozen or adaptive single-resolution recipe.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Differences, supplied-scale residuals, and all/any threshold conditions produce the declared derived values.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/compare-declared-metrics","title":"Find changes in conversion, unit cost, and metric gaps","criterion":"Missing sources and zero denominators remain unavailable with a reason rather than becoming zero.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/compare-declared-metrics --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","order":20}},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","status":"implemented","area":"relationships","path":".capabilities/relationships/comparisons/evaluate-delayed-response.capability.yaml","health":"review","coverage":"partial","summary":"A lagged-response relationship compares current response with a supplied or training-fitted multiple of an earlier predictor.","intent":"Inspect whether a metric still follows the delayed relationship declared by the caller.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"A supplied lag and coefficient use the predictor at the declared earlier sample position.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Derived values retain response and predictor source indexes and source cutoffs in lineage.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/comparisons/evaluate-delayed-response","title":"Check whether a metric follows its expected delayed response","criterion":"Samples lacking the earlier predictor or awaiting coefficient training remain unavailable.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/comparisons/evaluate-delayed-response --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Check delayed responses","order":30}},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","status":"implemented","area":"relationships","path":".capabilities/relationships/quality/require-compatible-sources.capability.yaml","health":"review","coverage":"partial","summary":"Declared relationships require exact UTC timestamp alignment, shared cadence and entity scope, and compatible declared units.","intent":"Prevent comparisons between mismatched populations or periods from creating misleading evidence.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"An invalid relationship is reported as failed while valid sibling dataset results remain available.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Difference relationships require identical declared source units and matching output units.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"relationships/quality/require-compatible-sources","title":"Compare the same entity and time period","criterion":"Related datasets use identical normalized timestamps, the same declared cadence, and matching entity metadata.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify relationships/quality/require-compatible-sources --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Related-metric analysis","backbone":"Investigate relationships I declare","step":"Align comparable sources","order":10}},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","status":"implemented","area":"relationships","path":".capabilities/relationships/resolution/compare-finalized-subperiods.capability.yaml","health":"review","coverage":"partial","summary":"Extend declared related-metric analysis to compatible finalized subperiods and completed fixed UTC 24-hour additive aggregates.","intent":"Detect conversion or unit-cost changes during a period while keeping partial progress separate from completed-period comparisons.","references":{"total":6,"readable":6,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":4,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":0,"ignored":0},"nextAction":"Run semantic review, save review evidence, or ignore accepted findings.","topFindings":[{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Callers select related datasets with identical native boundaries, entity scope, declared aggregation semantics, and compatible period definitions.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":22,"excerpt":"cases = compose_cases([d for d in datasets if d.resolution == resolution],"},{"reference":"src/anomalyzer/sparse_views.py","line":146,"excerpt":"native_sources = [d.model_copy(update={\"data_policy\": d.data_policy or DataPolicy()}) for d in request.datasets]"},{"reference":"src/anomalyzer/sparse_views.py","line":149,"excerpt":"native_request = request.model_copy(update={\"datasets\": native_sources, \"config\": native_config})"},{"reference":"src/anomalyzer/sparse_views.py","line":169,"excerpt":"daily_request = request.model_copy(update={\"datasets\": sources, \"config\": daily_config, \"relationships\": daily_relationships})"},{"reference":"src/anomalyzer/agent.py","line":14,"excerpt":"\"\"\"Use legacy semantics for a Dataset; explicit V11 fields select source policies.\"\"\""},{"reference":"src/anomalyzer/agent.py","line":20,"excerpt":"name in dataset for name in (\"data_policy\", \"coverage\", \"available_at\", \"revisions\", \"rules\", \"completed_period_rules\", \"event_ids\", \"event_sequence\")):"},{"reference":"src/anomalyzer/agent.py","line":23,"excerpt":"return analyze({\"schema_version\": \"1.1\", \"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"src/anomalyzer/agent.py","line":29,"excerpt":"return analyze({\"datasets\": [validated.model_dump(mode=\"json\")],"},{"reference":"tests/test_sparse.py","line":18,"excerpt":"return {\"schema_version\": \"1.1\", \"datasets\": [{\"id\": \"series\", \"entity\": {\"customer\": \"test\"},"},{"reference":"tests/test_sparse.py","line":60,"excerpt":"request[\"datasets\"][0][\"timestamps\"][2] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":88,"excerpt":"def test_declared_duplicate_aggregation_preserves_provenance(policy, value):"},{"reference":"tests/test_sparse.py","line":100,"excerpt":"request[\"datasets\"][0][\"timestamps\"][-1] = \"2026-01-01T03:00:00Z\""},{"reference":"tests/test_sparse.py","line":118,"excerpt":"changed[\"datasets\"][0][\"values\"][-10:] = [999]*10"},{"reference":"docs/sparse-data.md","line":36,"excerpt":"result = analyze({\"schema_version\": \"1.1\", \"datasets\": [source.model_dump(mode=\"json\")]})"},{"reference":"docs/sparse-data.md","line":117,"excerpt":"Schema 1.1 adds the optional **`event-identity-v1`** contract. It is selected by"},{"reference":"docs/sparse-data.md","line":118,"excerpt":"aligned `DatasetV11.event_ids` on an explicitly declared event-index source:"},{"reference":"docs/sparse-data.md","line":156,"excerpt":"`data_quality.event_identity_contract=\"event-identity-v1\"` record those semantics."},{"reference":"docs/sparse-data.md","line":206,"excerpt":"and a selected-input fingerprint. Evidence IDs use stable event identity within"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Completed aggregate ratios use aggregated numerators and denominators; a ratio of totals is not silently replaced by an average of individual ratios.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":29,"excerpt":"case.non_claims.append(\"Native and completed-day views reuse the same observations; they are not independent confirmation.\")"},{"reference":"src/anomalyzer/sparse_views.py","line":70,"excerpt":"\"\"\"Sum only wholly observed finalized additive children. Never average ratios.\"\"\""},{"reference":"src/anomalyzer/sparse_views.py","line":83,"excerpt":"unsupported = \"completed-day view requires additive_volume and sum; per-event percentages/ratios are not averaged\""},{"reference":"src/anomalyzer/sparse_views.py","line":175,"excerpt":"applicability=\"completed-day view supports ratio-of-sums and differences of compatible additive sums only\","},{"reference":"src/anomalyzer/relationships.py","line":94,"excerpt":"used = [(numerator, index), (denominator, index)]"},{"reference":"src/anomalyzer/relationships.py","line":177,"excerpt":"\"complete\" if source.values[i] is not None else \"unknown_or_incomplete\" for source, i in used}"},{"reference":"tests/test_sparse.py","line":231,"excerpt":"request[\"relationships\"] = [{\"id\": \"unit-cost\", \"kind\": \"ratio\", \"numerator\": \"series\", \"denominator\": \"calls\", \"units\": \"dollars_per_call\"}]"},{"reference":"tests/test_sparse.py","line":260,"excerpt":"def test_daily_rules_are_not_silently_reused_from_hourly_rules():"},{"reference":"tests/test_sparse.py","line":279,"excerpt":"def test_adaptive_gap_does_not_silently_train_a_post_gap_episode():"},{"reference":"tests/test_sparse.py","line":322,"excerpt":"def test_restart_policy_cannot_reuse_a_prior_regime_reference():"},{"reference":"tests/test_sparse.py","line":431,"excerpt":"assert row[\"complete\"] and row[\"aggregate_value\"] == 24 and not row[\"model_eligible\"]"},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":220,"excerpt":"relationship cannot use unordered tied events. Completed-day event-index"},{"reference":"docs/sparse-data.md","line":239,"excerpt":"It then collects `calibration_size` eligible forecast residuals, not elapsed slots."},{"reference":"docs/sparse-data.md","line":261,"excerpt":"reset or caller-selected `reference_gaps=\"restart\"`; the default does not silently"},{"reference":"docs/sparse-data.md","line":332,"excerpt":"partial sums are progress metadata, never a completed total. Empty days are not"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."},{"capabilityId":"relationships/resolution/compare-finalized-subperiods","title":"Inspect related metrics before the business period closes","criterion":"Finalized subperiod relationships retain each source value and original source index; unavailable or incomplete sources yield explicit unavailability.","evidence":[{"reference":"src/anomalyzer/sparse_views.py","line":19,"excerpt":"def view_cases(datasets, relationships, request):"},{"reference":"src/anomalyzer/sparse_views.py","line":76,"excerpt":"unsupported = \"completed-day aggregation is unavailable on event_index\""},{"reference":"src/anomalyzer/sparse_views.py","line":95,"excerpt":"raise ValueError(\"source timestamps are off the aggregate_anchor grid\")"},{"reference":"src/anomalyzer/sparse_views.py","line":129,"excerpt":"source_input_indexes=[raw for i in indexes for raw in quality[\"samples\"][i][\"source_indexes\"]],"},{"reference":"src/anomalyzer/sparse_views.py","line":130,"excerpt":"source_grid_indexes=indexes, observed_sum=(math.fsum(measurements[i] for i in indexes if measurements[i] is not None)"},{"reference":"src/anomalyzer/relationships.py","line":7,"excerpt":"from .alignment import PreparedDataset, exact_sources"},{"reference":"src/anomalyzer/relationships.py","line":19,"excerpt":"def _source_ids(relationship: Relationship) -> list[str]:"},{"reference":"src/anomalyzer/relationships.py","line":33,"excerpt":"ids = _source_ids(relationship)"},{"reference":"src/anomalyzer/relationships.py","line":34,"excerpt":"sources = exact_sources(ids, prepared)"},{"reference":"src/anomalyzer/relationships.py","line":35,"excerpt":"identified = \"event_identity_contract\" in sources[0].quality"},{"reference":"src/anomalyzer/agent.py","line":34,"excerpt":"def analyze_relationships(request: RequestV11 | dict) -> ResultV11:"},{"reference":"tests/test_sparse.py","line":10,"excerpt":"from anomalyzer import analyze, analyze_series, analyze_relationships, get_case, replay_policy"},{"reference":"tests/test_sparse.py","line":41,"excerpt":"assert right.data_quality[\"samples\"][8][\"source_index\"] == -1"},{"reference":"tests/test_sparse.py","line":51,"excerpt":"assert [r[\"original_value\"] for r in rows] == [2, None, None, 0, 9]"},{"reference":"tests/test_sparse.py","line":61,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, 0, 0, 2]"},{"reference":"tests/test_sparse.py","line":63,"excerpt":"assert [r[\"value\"] for r in source(request).data_quality[\"samples\"]] == [1, None, 0, 2]"},{"reference":"docs/sparse-data.md","line":22,"excerpt":"works through `analyze`, `analyze_relationships`, CLI JSON and read-only MCP."},{"reference":"docs/sparse-data.md","line":46,"excerpt":"| `coverage=\"incomplete\"` | Original partial value remains visible but is excluded from completed measurements, references, calibration and totals. |"},{"reference":"docs/sparse-data.md","line":73,"excerpt":"Preparation reports original values, model values, coverage, source indexes,"},{"reference":"docs/sparse-data.md","line":77,"excerpt":"For policy-prepared sources, `grid_count` counts positions, `observation_count`"},{"reference":"docs/sparse-data.md","line":107,"excerpt":"retains its unique-source requirement. With an explicit `data_policy`, complete,"}],"status":"weak-evidence","confidence":"medium","action":"inspect-evidence","rationale":"Related implementation text was found, but deterministic matching cannot prove the behavior.","recommendation":"Review the cited evidence. If it is correct, save agent.review evidence or make the criterion/reference wording more explicit."}],"storyMap":{"release":"Related-metric analysis","backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","order":30}}],"deliveryStrategy":{"release":"Related-metric analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["relationships/quality/require-compatible-sources","relationships/resolution/compare-finalized-subperiods"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"relationships/quality/require-compatible-sources needs review; relationships/resolution/compare-finalized-subperiods needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Investigate relationships I declare","Monitor the periods my business uses"],"missingBackbones":[],"stepCoverageGaps":["Investigate relationships I declare > Check delayed responses","Investigate relationships I declare > Compare ratios and metric gaps"],"rationale":"Starts with one step from each mapped backbone: Investigate relationships I declare, Monitor the periods my business uses."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["relationships/comparisons/compare-declared-metrics","relationships/comparisons/evaluate-delayed-response"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"relationships/comparisons/compare-declared-metrics needs review; relationships/comparisons/evaluate-delayed-response needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Investigate relationships I declare"],"missingBackbones":["Monitor the periods my business uses"],"stepCoverageGaps":["Investigate relationships I declare > Align comparable sources","Monitor the periods my business uses > Compare related metrics while the period is open"],"rationale":"Prioritizes capabilities with delivery risk signals: relationships/comparisons/compare-declared-metrics, relationships/comparisons/evaluate-delayed-response."}]},"presentation":{"outcome":"Related-metric analysis helps teams move from Align comparable sources to Compare related metrics while the period is open across 2 backbone activities.","narrativePath":[{"backbone":"Investigate relationships I declare","step":"Align comparable sources","capabilityIds":["relationships/quality/require-compatible-sources"],"health":"review"},{"backbone":"Investigate relationships I declare","step":"Compare ratios and metric gaps","capabilityIds":["relationships/comparisons/compare-declared-metrics"],"health":"review"},{"backbone":"Investigate relationships I declare","step":"Check delayed responses","capabilityIds":["relationships/comparisons/evaluate-delayed-response"],"health":"review"},{"backbone":"Monitor the periods my business uses","step":"Compare related metrics while the period is open","capabilityIds":["relationships/resolution/compare-finalized-subperiods"],"health":"review"}],"coverageSignals":[{"kind":"missing","label":"Monitor the periods my business uses","message":"Monitor the periods my business uses is not covered in the recommended slice."},{"kind":"weak","label":"Investigate relationships I declare > Align comparable sources","message":"Investigate relationships I declare > Align comparable sources has needs-review coverage."},{"kind":"weak","label":"Investigate relationships I declare > Check delayed responses","message":"Investigate relationships I declare > Check delayed responses has needs-review coverage."},{"kind":"weak","label":"Investigate relationships I declare > Compare ratios and metric gaps","message":"Investigate relationships I declare > Compare ratios and metric gaps has needs-review coverage."},{"kind":"weak","label":"Monitor the periods my business uses > Compare related metrics while the period is open","message":"Monitor the periods my business uses > Compare related metrics while the period is open has needs-review coverage."}]}},{"release":"Trend and seasonal analysis","capabilities":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","status":"implemented","area":"analysis","path":".capabilities/analysis/detection/identify-residual-patterns.capability.yaml","health":"review","coverage":"partial","summary":"Residual rules identify location shifts, trends, oscillation, mixture patterns, and short-term variation beyond isolated point departures.","intent":"Find accumulating changes and explain which numerical pattern was detected.","references":{"total":3,"readable":3,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"CUSUM reports the first detection position, direction, accumulated score, and configured threshold.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Moving-range findings identify variation changes separately from location changes.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/detection/identify-residual-patterns","title":"Spot sustained shifts and changing patterns","criterion":"Nelson location rules and CUSUM can identify a sustained residual location shift without any point-threshold crossing.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/detection/identify-residual-patterns --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","order":60}},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/adapt-to-changing-growth.capability.yaml","health":"review","coverage":"partial","summary":"The adaptive recipe updates each expectation from a bounded window of earlier observations while retaining frozen residual calibration.","intent":"Follow evolving growth without losing evidence of an isolated unusual value.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"An isolated extreme value is held out of future model references rather than creating a seasonal echo.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Conflicting trend choices and slope lookbacks exceeding the adaptive window are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/adapt-to-changing-growth","title":"Follow evolving growth while exposing unusual values","criterion":"Each rolling expectation uses only observations before the sample being scored.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/adapt-to-changing-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","order":40}},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/model-seasonal-level-and-growth.capability.yaml","health":"review","coverage":"partial","summary":"The default recipe builds an expectation from earlier matching seasonal observations and a training-fitted level, linear, or compound-growth model.","intent":"Distinguish unusual departures from ordinary repeating patterns and growth.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"A supplied season length selects earlier matching seasonal observations for the reference.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Automatic trend selection represents suitable linear and compound growth without flagging ordinary growth as point anomalies.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/model-seasonal-level-and-growth","title":"Account for ordinary seasonal patterns and growth","criterion":"Callers can explicitly select none, linear, or exponential trend, and unsupported trend modes are rejected.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/model-seasonal-level-and-growth --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","order":10}},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/protect-reference-from-extremes.capability.yaml","health":"review","coverage":"partial","summary":"Robust mode limits extreme observations' influence on trend fitting, residual calibration, and later seasonal references.","intent":"Keep one incident from inflating the reference or appearing again as a seasonal echo.","references":{"total":7,"readable":7,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"An extreme evaluated sample remains evidence while its expected value is used in future references.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Callers can choose include mode to retain raw extreme observations and inspect the resulting echo behavior.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/protect-reference-from-extremes","title":"Limit how extreme values distort later comparisons","criterion":"Robust seasonal references reduce the influence of an extreme prior matching seasonal value.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/protect-reference-from-extremes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","order":30}},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/reset-declared-regimes.capability.yaml","health":"review","coverage":"partial","summary":"Caller-supplied reset points divide a single-resolution series into independently trained and calibrated regimes.","intent":"Recognize known discontinuities without forcing the previous regime to explain the new one.","references":{"total":4,"readable":4,"missing":[]},"verification":{"automated":3,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"CLI reset options produce the same resolved positions and suppress departures explained by the declared boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Each new regime rebuilds its reference, trend, and calibration without crossing earlier boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/reset-declared-regimes","title":"Compare against a new baseline after a known change","criterion":"Integer positions and exact observed dates or timestamps can identify reset boundaries.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/reset-declared-regimes --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","order":50}},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","status":"implemented","area":"analysis","path":".capabilities/analysis/expectations/select-season-length.capability.yaml","health":"review","coverage":"partial","summary":"Callers can supply a season length; ordered values without one can use conservative season inference with a lag-one fallback.","intent":"Make seasonal assumptions explicit and inspectable for data without timestamps.","references":{"total":5,"readable":5,"missing":[]},"verification":{"automated":4,"manual":1,"gaps":[]},"counts":{"covered":0,"weak-evidence":0,"implementation-gap":0,"missing-reference":0,"no-implementation-reference":0,"assessor-limitation":4,"ignored":0},"nextAction":"Split broad criteria or run semantic review.","topFindings":[{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A positional series without a sufficiently stable candidate falls back to lag one and records why.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"A repeating positional series can infer a seasonal lag and report the selection window.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]},{"capabilityId":"analysis/expectations/select-season-length","title":"Choose the repeating cycle for a fair comparison","criterion":"An explicit season length is honored and reported as explicit.","status":"assessor-limitation","confidence":"low","action":"manual-review","rationale":"Saved review is stale: capability or implementation evidence changed.","recommendation":"Run capabilitykit verify analysis/expectations/select-season-length --agent <command> to refresh this capability.","evidence":[]}],"storyMap":{"release":"Trend and seasonal analysis","backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","order":20}}],"deliveryStrategy":{"release":"Trend and seasonal analysis","recommendations":[{"order":1,"phase":"opening","name":"Walking skeleton","capabilityIds":["analysis/expectations/model-seasonal-level-and-growth"],"releaseStrategy":"Prove one coherent end-to-end slice before deepening individual steps.","developmentStrategy":"Integrate the earliest capability from each backbone first, keeping scope thin until the narrative works.","riskIntent":"analysis/expectations/model-seasonal-level-and-growth needs review","learningIntent":"Validate whether users and stakeholders can recognize a meaningful outcome across the release backbone.","backboneCoverage":["Refine what ordinary behavior means"],"missingBackbones":[],"stepCoverageGaps":["Refine what ordinary behavior means > Choose a repeating cycle","Refine what ordinary behavior means > Find sustained residual changes","Refine what ordinary behavior means > Follow evolving growth","Refine what ordinary behavior means > Protect later comparisons","Refine what ordinary behavior means > Restart after a reviewed change"],"rationale":"Starts with the earliest mapped step in Refine what ordinary behavior means."},{"order":2,"phase":"end-game","name":"Risk burn-down and release hardening","capabilityIds":["analysis/expectations/select-season-length","analysis/expectations/protect-reference-from-extremes","analysis/expectations/adapt-to-changing-growth","analysis/expectations/reset-declared-regimes","analysis/detection/identify-residual-patterns"],"releaseStrategy":"Resolve delivery risks once the release story is visible enough to judge tradeoffs.","developmentStrategy":"Target review, action, verification, and reference gaps with explicit finish-line criteria.","riskIntent":"analysis/detection/identify-residual-patterns needs review; analysis/expectations/adapt-to-changing-growth needs review; analysis/expectations/protect-reference-from-extremes needs review; analysis/expectations/reset-declared-regimes needs review; analysis/expectations/select-season-length needs review","learningIntent":"Learn whether remaining risks change scope, release confidence, or stakeholder messaging.","backboneCoverage":["Refine what ordinary behavior means"],"missingBackbones":[],"stepCoverageGaps":["Refine what ordinary behavior means > Account for growth and seasonality"],"rationale":"Prioritizes capabilities with delivery risk signals: analysis/expectations/select-season-length, analysis/expectations/protect-reference-from-extremes, analysis/expectations/adapt-to-changing-growth, analysis/expectations/reset-declared-regimes, analysis/detection/identify-residual-patterns."}]},"presentation":{"outcome":"Trend and seasonal analysis helps teams move from Account for growth and seasonality to Find sustained residual changes across 1 backbone activity.","narrativePath":[{"backbone":"Refine what ordinary behavior means","step":"Account for growth and seasonality","capabilityIds":["analysis/expectations/model-seasonal-level-and-growth"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Choose a repeating cycle","capabilityIds":["analysis/expectations/select-season-length"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Protect later comparisons","capabilityIds":["analysis/expectations/protect-reference-from-extremes"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Follow evolving growth","capabilityIds":["analysis/expectations/adapt-to-changing-growth"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Restart after a reviewed change","capabilityIds":["analysis/expectations/reset-declared-regimes"],"health":"review"},{"backbone":"Refine what ordinary behavior means","step":"Find sustained residual changes","capabilityIds":["analysis/detection/identify-residual-patterns"],"health":"review"}],"coverageSignals":[{"kind":"weak","label":"Refine what ordinary behavior means > Account for growth and seasonality","message":"Refine what ordinary behavior means > Account for growth and seasonality has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Choose a repeating cycle","message":"Refine what ordinary behavior means > Choose a repeating cycle has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Find sustained residual changes","message":"Refine what ordinary behavior means > Find sustained residual changes has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Follow evolving growth","message":"Refine what ordinary behavior means > Follow evolving growth has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Protect later comparisons","message":"Refine what ordinary behavior means > Protect later comparisons has needs-review coverage."},{"kind":"weak","label":"Refine what ordinary behavior means > Restart after a reviewed change","message":"Refine what ordinary behavior means > Restart after a reviewed change has needs-review coverage."}]}}]},"summary":{"total":49,"ok":0,"review":32,"action":0,"planned":17}};
 let selectedRelease = "";
 let selectedHealth = "";
 let searchText = "";
diff --git a/README.md b/README.md
index c3b55c6..527f5fa 100644
--- a/README.md
+++ b/README.md
@@ -30,7 +30,7 @@ For application builds without a source checkout, use a pinned wheel from
 [GitHub Releases](https://github.com/PredictabilityAtScale/anomaly-detection/releases).
 The [release and library-consumer guide](docs/releases/README.md) explains the
 first release, dependency hashes, upgrades and Linux/Python 3.12 bundling for Guard.
-Version 0.2.1 includes the sparse-data progression in chapters 14–22. Tagged
+Version 0.3.0 includes the sparse-data progression in chapters 14–24. Tagged
 artifacts are published after the workflow passes and the owner seals the draft.
 
 Install from a checkout with Python 3.11–3.13:
@@ -193,7 +193,7 @@ when visualization is unavailable.
 Schema 1.0 is the single-series contract. Schema 1.1 adds named datasets,
 relationships, assessments, policy, and cases; it also works with one dataset
 and no relationships. Both use the same analysis core through CLI, Python, and
-MCP. Sparse schema-1.1 and multi-resolution support is described in chapters 14–22.
+MCP. Sparse schema-1.1, multi-resolution and identified events are described in chapters 14–24.
 The legacy schema-1.0 multi-resolution profile does
 not support `reset_points`; both other recipes support explicit resets.
 
@@ -730,6 +730,54 @@ the remaining complete-span restrictions. All new fixtures run through
 `examples/run_demo.py`; the existing generator and figure renderer reproduce the
 full progression.
 
+### 23. Keep distinct calls at the same recorded time
+
+Two completed calls can share the exact timestamp while having different costs.
+The [identified-call example](examples/sparse/identified_calls.json) supplies
+aligned `event_ids` under `data_policy.observation_basis="event_index"`.
+It retains all ten original calls: three in the training cohort, four in the
+calibration cohort, and three at `09:00:10.789Z`. The last three cost $0.80,
+$0.12 and $0.90. They remain three observations at the same time; no jitter,
+averaging or delivery order is introduced.
+
+![Distinct call costs at exact shared timestamps, with separate IDs and a frozen prior-time reference](docs/images/anomaly-progression-23-same-time-calls.png)
+
+The $0.80 and $0.90 calls each have inspectable point/range evidence and a distinct
+`Case.event_id`. A whole timestamp cohort finishes training or calibration before
+a later cohort can use that reference; alphabetical ID order never trains one
+same-time call on another. Actual support can therefore exceed the requested
+minimum training/calibration count. Event IDs and source indexes link evidence
+back to the original measurements. Separate findings do not imply separate
+notifications or independent confirmations.
+
+### 24. Declare identity, retries and order separately
+
+`event-identity-v1` requires caller-trusted logical IDs. A retry keeps the same
+ID; the [duplicate-delivery fixture](examples/sparse/identified_retry.json) is
+rejected instead of counted as another call. Repeating a valid snapshot preserves
+its case and evidence IDs. The [correction fixture](examples/sparse/identified_corrected.json)
+addresses `call-07` using `SampleRevision.event_id`, keeps its original timestamp,
+and changes its case revision while retaining the case ID. The
+[earlier as-of snapshot](examples/sparse/identified_before_correction.json) still
+returns the original evidence.
+
+![Identical same-time residuals support points but no timestamp-only sequence pattern; an explicit trusted sequence permits order-based checks](docs/images/anomaly-progression-24-event-order.png)
+
+In the [timestamp-only example](examples/sparse/identified_pattern_ties.json),
+twelve comparable evaluation calls share one timestamp. Their point reference
+is usable, but their ID layout cannot establish change-rule, Nelson, CUSUM or
+moving-range continuity. Each tied call breaks those comparisons. There is no
+sequence-pattern case in this example.
+
+The [ordered variant](examples/sparse/identified_pattern_sequence.json) adds
+`data_policy.event_order="trusted_sequence"` and unique aligned integer
+`event_sequence` ordinals. Only a caller with an actual trustworthy ordering
+contract should assert that order. The same timestamps stay unchanged; the
+declared sequence permits residual pattern checks without establishing elapsed
+rates, causal explanation or independence. Without IDs, the existing unique-time
+event-index contract continues unchanged. See the [API guide](docs/sparse-data.md#stable-event-identities-since-030)
+for validation, replay, alignment and caller responsibilities.
+
 ## Baseline and calibration reference
 
 The robust expectation for sample `t` uses four previous matching seasonal
diff --git a/docs/images/anomaly-progression-23-same-time-calls.png b/docs/images/anomaly-progression-23-same-time-calls.png
new file mode 100644
index 0000000..3f6beee
Binary files /dev/null and b/docs/images/anomaly-progression-23-same-time-calls.png differ
diff --git a/docs/images/anomaly-progression-24-event-order.png b/docs/images/anomaly-progression-24-event-order.png
new file mode 100644
index 0000000..3a273be
Binary files /dev/null and b/docs/images/anomaly-progression-24-event-order.png differ
diff --git a/docs/releases/0.3.0.md b/docs/releases/0.3.0.md
new file mode 100644
index 0000000..c9009f3
--- /dev/null
+++ b/docs/releases/0.3.0.md
@@ -0,0 +1,48 @@
+# Anomalyzer 0.3.0 — distinct events at the same original time
+
+Schema 1.1 adds the optional `event-identity-v1` contract for comparable event-index
+measurements. `DatasetV11.event_ids` retain distinct same-time calls without
+jitter, aggregation or lost values. Repeated logical IDs are rejected, including
+identical retry deliveries; callers own trustworthy identity and upstream deduplication.
+
+- Default `data_policy.event_order="timestamp"` sorts time/ID for layout only.
+  Whole timestamp cohorts finish training/calibration before a later timestamp
+  uses the reference. Tied observations keep point and explicit range evidence,
+  break change comparisons, and cannot create sequence-pattern continuity.
+- Optional `event_order="trusted_sequence"` requires aligned unique integer
+  `event_sequence` ordinals consistent with nondecreasing original times. This
+  is a caller assertion of order, not independence, causation or elapsed cadence.
+- `SampleRevision.event_id` addresses a correction to one existing event at its
+  original time. Availability/as-of selection is deterministic. Case and evidence
+  IDs use logical identity; relevant changes create a new case revision. Cases,
+  point episodes and rule findings can overlap and are not alert counts.
+- Assessments/cases expose `event_id`; quality samples preserve original time,
+  original input index, initial value and selected revision provenance. MCP
+  `get_evidence.source_samples` makes distinct same-time calls inspectable.
+- Exact identified relationships require matching logical identities and order
+  contracts. Lineage exposes source event IDs and original timestamp strings.
+  Unordered tied events cannot feed a lagged relationship.
+- README chapters 23–24 continue the existing numbered progression with six
+  reproducible fixtures and two rendered charts. Source tests cover schemas,
+  Python, CLI, MCP, prefix/as-of/corrections and bounded result growth. Installed
+  artifact checks and the offline Linux x86_64/Python 3.12 consumer exercise IDs.
+- The standalone Linux benchmark includes 1,344 distinct calls with four per
+  timestamp. Its result sizes and warm timings are mechanics checks, not measured
+  production capacity, accuracy, memory consumption or cold-start cost.
+- Release staging retries reads after draft creation to tolerate GitHub's draft
+  index delay without creating or overwriting another release.
+
+Calendar grids and legacy unique-time event inputs keep their existing numerical
+semantics; schema 1.0 remains unchanged. Upgrade strict schema-1.1 readers/producers
+together for the added fields. The package remains alpha, supports Python 3.11–3.13,
+and is distributed as an immutable GitHub wheel/sdist with hash-pinned dependencies.
+
+Limits: the engine cannot authenticate event IDs/order, detect aliases, maintain
+a durable retry/arrival ledger, infer independence, or finalize an open same-time
+cohort. Adding a newly available member/correction can revise later references;
+explicit availability preserves earlier as-of snapshots. Event-index calendar
+aggregation, multi-seasonality and business incident/notification policy remain
+outside this contract. This release does not change or deploy UsageTap Guard.
+
+See `docs/sparse-data.md`, README chapters 23–24 and `docs/releases/README.md` for
+the exact API, replay boundaries, examples and Linux build-time installation.
diff --git a/docs/releases/README.md b/docs/releases/README.md
index 1520d98..23e3539 100644
--- a/docs/releases/README.md
+++ b/docs/releases/README.md
@@ -1,13 +1,13 @@
 # Release and use Anomalyzer
 
-Anomalyzer is a normal Python dependency. Version **0.2.1** is the first published
+Anomalyzer is a normal Python dependency. Version **0.3.0** is the current
 release; nothing is uploaded to PyPI by this workflow. GitHub Releases is the
 initial distribution channel. Publishing the library does not integrate it into
 UsageTap Guard.
 
 The original `v0.2.0` tag remains an unpublished candidate after a draft-lookup
 failure. Its tag and draft are retained; the fixed publication helper ships in
-`v0.2.1`.
+`v0.2.1`; that was the first published release.
 
 ## Four terms
 
@@ -15,7 +15,7 @@ failure. Its tag and draft are retained; the fixed publication helper ships in
 | --- | --- |
 | Wheel (`.whl`) | Ready-to-install Python package. Use this for Guard. Anomalyzer's wheel works across supported operating systems; its dependencies may need platform-specific wheels. |
 | Source distribution (`.tar.gz`, or sdist) | Source needed to build the package. Useful for rebuilding and inspection; installation also needs the declared build backend. |
-| Version tag (`v0.2.1`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |
+| Version tag (`v0.3.0`) | Permanent Git name for the exact source revision released. It must match `_version.py`; never move or reuse it. |
 | Dependency pin | An exact version or artifact URL, plus a SHA-256 fingerprint. A hash checks that downloaded bytes match the intended artifact. Pin transitive dependencies too. |
 
 The package supports Python 3.11, 3.12 and 3.13. Version lives only in
@@ -45,7 +45,7 @@ it for package metadata and the public `__version__` imports it.
 Before choosing a version, inspect existing tags/releases and the immutability
 setting. Never reuse a published version or move a version tag.
 
-## First release: 0.2.1
+## Release 0.3.0
 
 Run these commands from a clean checkout of the merged `main` branch. Commands
 work in PowerShell or Bash except where explicitly labelled.
@@ -61,16 +61,16 @@ uv sync --locked --extra dev
 uv run pytest -q
 uv run python examples/run_demo.py
 uv run python examples/run_agentic_demo.py
-uv run python scripts/release.py check-version --tag v0.2.1
+uv run python scripts/release.py check-version --tag v0.3.0
 ```
 
-`git status --short` must be empty. If `v0.2.1` already exists, stop and choose
+`git status --short` must be empty. If `v0.3.0` already exists, stop and choose
 the next unused version through a PR. Otherwise create an annotated tag on the
 reviewed commit and push it (do not use `--force`):
 
 ```text
-git tag -a v0.2.1 -m "Anomalyzer 0.2.1"
-git push origin v0.2.1
+git tag -a v0.3.0 -m "Anomalyzer 0.3.0"
+git push origin v0.3.0
 ```
 
 Open the repository's **Actions → Release** run. It runs all tests on
@@ -82,10 +82,10 @@ passes does it create a draft and attach the tested files. Publication is one
 explicit owner step, using the already tested files:
 
 ```text
-git switch --detach v0.2.1
-gh release download v0.2.1 --repo PredictabilityAtScale/anomaly-detection --dir release-v0.2.1
+git switch --detach v0.3.0
+gh release download v0.3.0 --repo PredictabilityAtScale/anomaly-detection --dir release-v0.3.0
 git rev-parse HEAD
-python scripts/publish_release.py --dist release-v0.2.1 --tag v0.2.1 --revision <full-SHA-printed-above> --publish
+python scripts/publish_release.py --dist release-v0.3.0 --tag v0.3.0 --revision <full-SHA-printed-above> --publish
 git switch main
 ```
 
@@ -104,14 +104,14 @@ any release asset.
 Assets live under [GitHub Releases](https://github.com/PredictabilityAtScale/anomaly-detection/releases),
 in the release for the selected tag:
 
-- `anomalyzer-0.2.1-py3-none-any.whl` and `anomalyzer-0.2.1.tar.gz`.
+- `anomalyzer-0.3.0-py3-none-any.whl` and `anomalyzer-0.3.0.tar.gz`.
 - `release-manifest.json`: version, tag, full source commit, file sizes and hashes.
 - `SHA256SUMS`: hashes of all assets including the manifest (except itself).
 - `runtime-dependencies.txt`: exact dependency versions and accepted hashes
   exported from the repository lock, including platform markers.
 - `requirements-guard-py312-linux-x86_64.txt`: those dependencies plus the exact
   GitHub wheel URL and its hash. This is an installable starting lock for Guard.
-- `requirements-offline.txt`: the same pins using `anomalyzer==0.2.1` so pip can
+- `requirements-offline.txt`: the same pins using `anomalyzer==0.3.0` so pip can
   install entirely from a local wheelhouse.
 
 CI's **Artifacts** section also holds `release-dist-<commit>` and the
@@ -143,7 +143,7 @@ release's bytes differ, leave it untouched and create a new version.
 
 Create a branch/PR, change only `_version.py` to the next unused `X.Y.Z` version,
 run `uv lock`, document changes in the PR, and run the same checks. Merge, then
-follow the first-release steps with the new tag. The package is alpha: review
+follow the release steps with the new tag. The package is alpha: review
 behavior and contracts before upgrading even a minor version. Avoid compatibility
 promises that the current alpha API cannot support.
 
@@ -156,11 +156,11 @@ analysis needs no checkout, CLI, MCP server, hosted Anomalyzer service or networ
 1. Download the chosen Release's assets and verify `SHA256SUMS`. For example:
 
    ```text
-   gh release download v0.2.1 --repo PredictabilityAtScale/anomaly-detection --dir release
+   gh release download v0.3.0 --repo PredictabilityAtScale/anomaly-detection --dir release
    ```
 
    In Linux/WSL/Bash: `(cd release && sha256sum --check SHA256SUMS)`.
-   In PowerShell, `Get-FileHash release/anomalyzer-0.2.1-py3-none-any.whl -Algorithm SHA256`
+   In PowerShell, `Get-FileHash release/anomalyzer-0.3.0-py3-none-any.whl -Algorithm SHA256`
    shows the wheel's hash; compare it with the manifest/checksum file. Also check
    the manifest's tag and commit against the intended release. Hashes identify
    bytes; use the trusted GitHub repository/release as their source.
@@ -169,7 +169,7 @@ analysis needs no checkout, CLI, MCP server, hosted Anomalyzer service or networ
    inputs. Its final requirement has this form, with a real generated hash:
 
    ```text
-   anomalyzer @ https://github.com/PredictabilityAtScale/anomaly-detection/releases/download/v0.2.1/anomalyzer-0.2.1-py3-none-any.whl --hash=sha256:<generated-wheel-hash>
+   anomalyzer @ https://github.com/PredictabilityAtScale/anomaly-detection/releases/download/v0.3.0/anomalyzer-0.3.0-py3-none-any.whl --hash=sha256:<generated-wheel-hash>
    ```
 
    Install the actual file, not this placeholder. It also pins Pydantic,
diff --git a/docs/sparse-data.md b/docs/sparse-data.md
index 46d52ea..2834300 100644
--- a/docs/sparse-data.md
+++ b/docs/sparse-data.md
@@ -1,16 +1,17 @@
-# Sparse observations in Anomalyzer 0.2
+# Sparse observations in Anomalyzer 0.3
 
 An empty hour does not tell us whether nothing happened or collection failed.
 Schema 1.1 therefore preserves missing values by default. An expectation can be
 shown separately; it is never inserted as an observed measurement. Schema 1.0
 keeps its existing complete-series behavior and unchanged schemas.
 Schema 1.1 gains optional source and result fields. Upgrade strict schema-1.1
-producers/readers together to 0.2; an older strict result model is not promised to
-accept the new metadata. Complete legacy requests remain accepted. Record the
+producers/readers together to 0.3 for the extended identity contract; an older
+strict result model is not promised to accept the new metadata. Complete legacy
+requests remain accepted. Record the
 package version and method `profile_revision` with Guard's own profile revision.
 
 The numbered [README progression](../README.md#14-choose-what-an-empty-period-means)
-continues with chapters 14–22. Fixtures live in [examples/sparse](../examples/sparse/).
+continues with chapters 14–24. Fixtures live in [examples/sparse](../examples/sparse/).
 Run `python examples/generate.py`, `python examples/run_demo.py` and
 `python examples/render_readme_progression.py --only-sparse` to reproduce them.
 The examples demonstrate mechanics, not measured abuse accuracy.
@@ -107,8 +108,126 @@ retains its unique-source requirement. With an explicit `data_policy`, complete,
 unrevisioned calendar duplicates can follow the declared `duplicate_policy`
 (`sum` or `mean`); original values/indexes remain in `duplicate_sources`. Nulls
 propagate. Duplicates involving incomplete coverage or revision metadata are
-rejected. Event-index timestamps must be unique; ties are refused rather than
-assigned invented times or counted as multiple independent events.
+rejected. Event-index timestamps without `event_ids` must be unique; identified
+events follow the separate contract below. Calendar duplicate policies do not
+deduplicate logical events.
+
+## Stable event identities since 0.3.0
+
+Schema 1.1 adds the optional **`event-identity-v1`** contract. It is selected by
+aligned `DatasetV11.event_ids` on an explicitly declared event-index source:
+
+```python
+from anomalyzer import analyze_series, DataPolicy, DatasetV11, SampleRevision
+
+calls = DatasetV11(
+    id="call-cost", entity={"application": "synthetic"}, units="dollars_per_call",
+    timestamps=["2026-01-01T09:00:10.789Z", "2026-01-01T09:00:10.789Z"],
+    values=[0.80, 0.12], event_ids=["call-A", "call-B"],
+    data_policy=DataPolicy(observation_basis="event_index"),
+    rules=[{"id": "illustrative-cap", "kind": "acceptable_range", "minimum": 0, "maximum": 0.20}],
+)
+result = analyze_series(calls, {"season_length": 1, "trend": "none"})
+assert result.cases[0].event_id == "call-A"  # range rule; no learned baseline yet
+```
+
+| Field | Contract |
+| --- | --- |
+| `event_ids` | Optional array aligned with original values/timestamps; every ID is a unique, case-sensitive string of 1–128 characters with no surrounding whitespace or line breaks. IDs require explicit event-index policy, original timestamps and no frequency. |
+| `data_policy.event_order` | `"timestamp"` by default. Sort by normalized time then exact ID for deterministic layout; equal-time ID order is **not** temporal order. |
+| `event_sequence` | Optional aligned array of unique strict integers in 0…9223372036854775807. Requires `event_order="trusted_sequence"` and IDs. Ordinals must agree with nondecreasing original timestamps. Gaps in ordinals are allowed and do not represent missing events or elapsed time. |
+| `SampleRevision.event_id` | Required for every revision of an identified source. It addresses an existing ID at exactly the same original timestamp. A revision changes value/coverage, never event identity, time or sequence. |
+
+The caller owns ID authenticity and namespace, retry deduplication and the
+comparability of measurements across customer/application/model/workload scope.
+The library cannot detect two aliases for the same call, prove a supplied ID is
+authentic, or establish independence from distinct IDs. Two deliveries of one
+logical event must keep one ID. Duplicate IDs in a request are rejected even if
+the delivered bytes match; deduplicate upstream. Across datasets, use the same ID
+only for measurements of the same logical event in the same declared entity.
+
+Without trusted sequence, a timestamp cohort cannot train or calibrate its own
+members. The engine admits the **whole eligible cohort** to the current training
+or calibration stage, so actual counts can exceed the configured minimum. A later
+timestamp uses the resulting frozen reference. All evaluation calls at the same
+time receive point comparisons against the same prior-time reference. Calibration
+cohorts are reference-only; this profile does not invent provisional within-cohort
+scores. `profile_revision="identified-events-v1"` and
+`data_quality.event_identity_contract="event-identity-v1"` record those semantics.
+
+An unordered tied cohort breaks change rules and descriptive prior populations
+before each tied member and immediately afterwards. Per-event range rules and
+appropriate point evidence remain usable. Tied members are excluded from Nelson,
+CUSUM, moving-range and adjacent-point runs; the remaining sequences break at
+their positions. Identified sequence patterns require at least two ordered
+evaluation points, so a lone large residual cannot masquerade as an accumulated
+sequence. Point episodes are per event and can overlap its case and range evidence.
+
+Only assert an actual trusted sequence when the source establishes event order:
+
+```python
+ordered = calls.model_copy(update={
+    "event_sequence": [410, 411],
+    "data_policy": DataPolicy(observation_basis="event_index", event_order="trusted_sequence"),
+})
+```
+
+This allows order-based references and sequence checks even when clock timestamps
+tie. It does not establish causation, statistical independence or elapsed rates.
+An integer reset may split trusted sequence; an unordered timestamp reset starts
+the whole cohort and a positional reset inside it is rejected. `reference_gaps="restart"`
+starts after an incomplete cohort rather than splitting its unordered members.
+
+Availability filters events **before** their coordinate/reference is constructed.
+Unannotated arrivals remain event-time snapshots, not known delivery history.
+Revisions must arrive strictly after the initial observation's supplied availability
+(or event time when absent); event/availability pairs and revision IDs are unique
+per logical event. The latest eligible revision wins independently of input order:
+
+```python
+correction = SampleRevision(event_id="call-A", timestamp=calls.timestamps[0],
+    value=0.85, available_at="2026-01-01T09:01:00Z", revision_id="corrected-cost-1")
+corrected = calls.model_copy(update={"revisions": [correction]})
+earlier = analyze_series(corrected, context={"as_of": "2026-01-01T09:00:30Z"})
+later = analyze_series(corrected, context={"as_of": "2026-01-01T09:02:00Z"})
+```
+
+Repeated valid snapshots and later-timestamp prefix extensions preserve prior
+point evidence. Adding another member to an **open same-time cohort** can change
+its available support and the interpretation of whether an order comparison was
+possible; it is not a finalized chronological prefix. Late arrivals with explicit
+availability never change an earlier as-of snapshot. A later snapshot may revise
+references after a late arrival/correction; persist the request and availability
+history if historical reconstruction is required. The engine maintains no durable
+delivery ledger or cross-request duplicate registry.
+
+Quality `samples` retain `event_id`, `original_timestamp`, `event_sequence`,
+original input indexes, `initial_value`, selected value, availability, revision ID
+and a selected-input fingerprint. Evidence IDs use stable event identity within
+the declared entity/target scope rather than array position; the existing `index`
+still links into prepared samples.
+`Assessment.event_id` and `Case.event_id` identify the call. A correction keeps the
+case ID while changing its revision when its numerical support changes. Entity
+scope and contributing targets also form case identity. Caller policy still owns
+notification/incident deduplication; a case, point episode and rule finding can be
+overlapping descriptions of the same call.
+
+CLI JSON and MCP accept the same fields. MCP case findings expose `event_id`;
+`get_evidence.source_samples` returns bounded source identity/provenance beside
+the selected evidence. Exact relationships between identified datasets require
+matching ID arrays, original UTC times and order contracts, with
+`LineagePoint.source_event_ids` and `source_original_timestamps`. A lagged
+relationship cannot use unordered tied events. Completed-day event-index
+aggregation remains unavailable. Calendar and legacy unique-time inputs keep
+their existing behavior; IDs cannot be used to bypass a calendar duplicate rule.
+
+Raw events plus revisions share the request point budget, and preparation/scoring
+use the cooperative runtime budget. CLI input reads enforce `max_bytes`;
+in-memory Python/MCP callers own ingress payload limits. Results grow with
+retained events and evidence; point limits are not a
+small-response guarantee. MCP summaries stay compact and drilldowns are paged.
+The standalone benchmark includes a 1,344-event workload with four distinct calls
+at each recorded timestamp; CI records actual result bytes and warm timings.
 
 ## What sparse statistical support means
 
diff --git a/examples/README.md b/examples/README.md
index 5d3b220..49c5997 100644
--- a/examples/README.md
+++ b/examples/README.md
@@ -100,3 +100,9 @@ Run a single case through the CLI:
 CSV and JSON contain the same values. Regenerate all twelve cases with
 `.venv/Scripts/python examples/generate.py`. Synthetic cases test mechanics;
 they do not establish accuracy on real usage data.
+
+Progression chapters 23–24 add `sparse/identified_calls.json`, the deliberately
+invalid retry fixture, addressed corrections/as-of snapshots, and timestamp-only
+versus explicitly trusted sequence examples. `examples/run_demo.py` checks every
+valid example and verifies that the retry is rejected. Render the complete sparse
+progression with `examples/render_readme_progression.py --only-sparse`.
diff --git a/examples/library/benchmark_sparse.py b/examples/library/benchmark_sparse.py
index 2d241bf..0dbc47e 100644
--- a/examples/library/benchmark_sparse.py
+++ b/examples/library/benchmark_sparse.py
@@ -38,6 +38,13 @@ def workloads():
         data_policy={"observation_basis":"event_index"})
     events["datasets"][0].pop("frequency")
     result["448_comparable_events"] = events
+    identified = deepcopy(base)
+    identified["datasets"][0].update(timestamps=[times[i//4] for i in range(count)],
+        values=[.10+(i%5)*.001 for i in range(count)], units="dollars_per_call",
+        event_ids=[f"call-{i:04}" for i in range(count)],
+        data_policy={"observation_basis":"event_index"})
+    identified["datasets"][0].pop("frequency")
+    result["1344_distinct_calls_with_timestamp_ties"] = identified
     return result
 
 
diff --git a/examples/library/guard_smoke.py b/examples/library/guard_smoke.py
index debc5c4..6dc5151 100644
--- a/examples/library/guard_smoke.py
+++ b/examples/library/guard_smoke.py
@@ -118,6 +118,16 @@ def offline(*args, **kwargs):
     events = {**sparse["datasets"][0], "data_policy":{"observation_basis":"event_index"}}
     events.pop("frequency")
     assert analyze_series(events, sparse["config"]).dataset_results[0].data_quality["coordinate"] == "event_index"
+    identified = dict(id="call-cost", entity={"application":"synthetic"}, units="dollars_per_call",
+        timestamps=["2026-01-01T00:00:00Z"]*2+["2026-01-01T01:00:00Z"]*3+["2026-01-01T02:00:00Z"]*2,
+        values=[.10,.10,.09,.10,.11,.80,.90], event_ids=[f"call-{i}" for i in range(7)],
+        data_policy={"observation_basis":"event_index"},
+        rules=[{"id":"example-cap","kind":"acceptable_range","minimum":0,"maximum":.20}])
+    actual = analyze_series(identified, {"season_length":1,"training_size":2,"calibration_size":3,"trend":"none"})
+    assert {case.event_id for case in actual.cases} == {"call-5","call-6"}
+    assert not actual.dataset_results[0].anomaly_patterns
+    assert actual.dataset_results[0].data_quality["event_identity_contract"] == "event-identity-v1"
+    assert [case.id for case in analyze_series(identified, {"training_size":2,"calibration_size":3,"trend":"none"}).cases] == [case.id for case in actual.cases]
     print(json.dumps({"version": anomalyzer.__version__, "status": result.status,
                       "cases": len(result.cases), "offline": True,
                       "pydantic_core": pydantic_core.__file__}))
diff --git a/examples/render_readme_progression.py b/examples/render_readme_progression.py
index 1a0eebf..7d9be82 100644
--- a/examples/render_readme_progression.py
+++ b/examples/render_readme_progression.py
@@ -988,6 +988,54 @@ def footer(svg, text, zeros=False):
     svg.text(70, 625, "Hourly evidence: available. Completed-day learning: insufficient history. Views share the same source and are correlated.", 17, INK)
     svg.text(70, 660, "Fixed UTC 24-hour windows; local-calendar/DST aggregation and adaptive multi-resolution children remain separate work.", 15, MUTED)
     figures["anomaly-progression-22-sparse-days"] = svg.finish()
+    svg = Svg()
+    header(svg, "23", "Keep distinct calls at the same recorded time", "Each blue or red mark is a real call at its exact timestamp; equal costs can overlap. IDs do not invent order.")
+    x_at = {"2026-01-01T09:00:00.123Z":180, "2026-01-01T09:00:05.456Z":630, "2026-01-01T09:00:10.789Z":1080}
+    sy = lambda value: 500-value/1.05*300
+    for cost in (0,.1,.2,.5,.8,1.0):
+        svg.line(95, sy(cost), 1130, sy(cost), GRID)
+        svg.text(80, sy(cost)+5, f"${cost:.2f}", 15, MUTED, anchor="end")
+    svg.line(95, 200, 95, 500, AXIS)
+    svg.line(95, 500, 1130, 500, AXIS)
+    svg.line(95, sy(.2), 1130, sy(.2), GOLD, 2, "6 5")
+    svg.text(110, sy(.2)-12, "Illustrative range maximum: $0.20", 16, GOLD)
+    svg.line(850, sy(.1), 1130, sy(.1), NAVY, 2, "7 5")
+    calls = requests["identified_calls"]["datasets"][0]
+    for time_text, cost, event_id in zip(calls["timestamps"], calls["values"], calls["event_ids"]):
+        svg.circle(x_at[time_text], sy(cost), 6, RED if cost > .2 else BLUE)
+    for text, x in x_at.items():
+        svg.text(x, 535, text[11:23], 16, MUTED, anchor="middle")
+    svg.text(180, 170, "3 training calls", 19, INK, 700, "middle")
+    svg.text(630, 170, "4 calibration calls", 19, INK, 700, "middle")
+    svg.text(1080, 170, "3 evaluated calls", 19, INK, 700, "middle")
+    svg.text(180, 560, "call-00 to call-02", 15, MUTED, anchor="middle")
+    svg.text(630, 560, "call-03 to call-06", 15, MUTED, anchor="middle")
+    svg.text(1080, 560, "call-07 to call-09", 15, MUTED, anchor="middle")
+    svg.text(1060, sy(.8)+5, "call-07: $0.80", 17, RED, 700, "end")
+    svg.text(1060, sy(.9)-10, "call-09: $0.90", 17, RED, 700, "end")
+    svg.text(1055, sy(.12)-13, "call-08: $0.12", 16, BLUE, 400, "end")
+    svg.text(60, 605, "Original UTC time (including milliseconds). Whole cohorts finish a reference stage before the next timestamp uses it.", 16, INK)
+    legend(svg, [("dot",BLUE,"measured call"),("dot",RED,"point / range criterion"),("dash",NAVY,"frozen earlier-time reference")], y=640)
+    svg.text(60, 669, "Two distinct call cases remain inspectable. Cases, rule findings and point episodes are overlapping evidence, not alert counts.", 14, MUTED)
+    figures["anomaly-progression-23-same-time-calls"] = svg.finish()
+    svg = Svg()
+    header(svg, "24", "Declare identity, retries and order separately", "Same values, IDs and timestamps in both panels; only the caller's explicit ordering contract changes.")
+    for name, y, title in (("identified_pattern_ties",175,"Timestamp-only: ID layout; tied calls supply no sequence pattern"),
+                           ("identified_pattern_sequence",410,"Trusted sequence: ordered residual checks become applicable")):
+        item = target(name)
+        values = observed(name)
+        displayed = values if name.endswith("sequence") else [None]*len(values)
+        sx, sy = chart(svg, displayed, [(NAVY,expectation(name),2,"7 5",1)],95,y,1035,135,
+            shade_until=None, y_bounds=(.08,.13), y_tick_format=lambda value:f"${value:.2f}",
+            ticks=[(0,"call-00"),(7,"call-07"),(18,"call-18")],label=title)
+        for i, cost in enumerate(values):
+            svg.circle(sx(i),sy(cost),3.5,BLUE)
+        svg.line(sx(7),y-5,sx(7),y+140,AXIS,1,"3 5")
+        svg.text(1120,y-17,f"{len(item.anomaly_patterns)} pattern finding(s)",16,NAVY,700,"end")
+    svg.text(95,585,"All twelve evaluation calls share 09:00:10.789Z. The x-axis is ID layout or declared ordinal, never elapsed time.",16,INK)
+    svg.text(60,630,"Retry: same logical ID is rejected twice. Correction: same event/case ID, addressed value revision, as-of history retained.",16,INK)
+    svg.text(60,665,"Only assert trusted_sequence from actual source order. Neither sequence nor distinct IDs proves independence or cause.",15,MUTED)
+    figures["anomaly-progression-24-event-order"] = svg.finish()
     return figures
 
 
diff --git a/examples/sparse/identified_before_correction.json b/examples/sparse/identified_before_correction.json
new file mode 100644
index 0000000..84b50df
--- /dev/null
+++ b/examples/sparse/identified_before_correction.json
@@ -0,0 +1,77 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.8,
+        0.12,
+        0.9
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ],
+      "revisions": [
+        {
+          "event_id": "call-07",
+          "timestamp": "2026-01-01T09:00:10.789Z",
+          "value": 0.85,
+          "available_at": "2026-01-01T09:01:00Z",
+          "revision_id": "corrected-cost-1"
+        }
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none"
+  },
+  "context": {
+    "as_of": "2026-01-01T09:00:30Z"
+  }
+}
diff --git a/examples/sparse/identified_calls.json b/examples/sparse/identified_calls.json
new file mode 100644
index 0000000..0038e32
--- /dev/null
+++ b/examples/sparse/identified_calls.json
@@ -0,0 +1,65 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.8,
+        0.12,
+        0.9
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none"
+  }
+}
diff --git a/examples/sparse/identified_corrected.json b/examples/sparse/identified_corrected.json
new file mode 100644
index 0000000..3fe9219
--- /dev/null
+++ b/examples/sparse/identified_corrected.json
@@ -0,0 +1,77 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.8,
+        0.12,
+        0.9
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ],
+      "revisions": [
+        {
+          "event_id": "call-07",
+          "timestamp": "2026-01-01T09:00:10.789Z",
+          "value": 0.85,
+          "available_at": "2026-01-01T09:01:00Z",
+          "revision_id": "corrected-cost-1"
+        }
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none"
+  },
+  "context": {
+    "as_of": "2026-01-01T09:02:00Z"
+  }
+}
diff --git a/examples/sparse/identified_pattern_sequence.json b/examples/sparse/identified_pattern_sequence.json
new file mode 100644
index 0000000..eaf9036
--- /dev/null
+++ b/examples/sparse/identified_pattern_sequence.json
@@ -0,0 +1,115 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09",
+        "call-10",
+        "call-11",
+        "call-12",
+        "call-13",
+        "call-14",
+        "call-15",
+        "call-16",
+        "call-17",
+        "call-18"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index",
+        "event_order": "trusted_sequence"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ],
+      "event_sequence": [
+        100,
+        101,
+        102,
+        103,
+        104,
+        105,
+        106,
+        107,
+        108,
+        109,
+        110,
+        111,
+        112,
+        113,
+        114,
+        115,
+        116,
+        117,
+        118
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none",
+    "point_threshold": 100
+  }
+}
diff --git a/examples/sparse/identified_pattern_ties.json b/examples/sparse/identified_pattern_ties.json
new file mode 100644
index 0000000..f93ba04
--- /dev/null
+++ b/examples/sparse/identified_pattern_ties.json
@@ -0,0 +1,93 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12,
+        0.12
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09",
+        "call-10",
+        "call-11",
+        "call-12",
+        "call-13",
+        "call-14",
+        "call-15",
+        "call-16",
+        "call-17",
+        "call-18"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none",
+    "point_threshold": 100
+  }
+}
diff --git a/examples/sparse/identified_retry.json b/examples/sparse/identified_retry.json
new file mode 100644
index 0000000..900a5c9
--- /dev/null
+++ b/examples/sparse/identified_retry.json
@@ -0,0 +1,68 @@
+{
+  "schema_version": "1.1",
+  "datasets": [
+    {
+      "id": "call-cost",
+      "entity": {
+        "application": "synthetic"
+      },
+      "units": "dollars_per_call",
+      "timestamps": [
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:00.123Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:05.456Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z",
+        "2026-01-01T09:00:10.789Z"
+      ],
+      "values": [
+        0.1,
+        0.1,
+        0.1,
+        0.09,
+        0.1,
+        0.11,
+        0.1,
+        0.8,
+        0.12,
+        0.9,
+        0.9
+      ],
+      "event_ids": [
+        "call-00",
+        "call-01",
+        "call-02",
+        "call-03",
+        "call-04",
+        "call-05",
+        "call-06",
+        "call-07",
+        "call-08",
+        "call-09",
+        "call-09"
+      ],
+      "data_policy": {
+        "observation_basis": "event_index"
+      },
+      "rules": [
+        {
+          "id": "illustrative-cap",
+          "kind": "acceptable_range",
+          "minimum": 0,
+          "maximum": 0.2
+        }
+      ]
+    }
+  ],
+  "config": {
+    "season_length": 1,
+    "training_size": 2,
+    "calibration_size": 3,
+    "trend": "none"
+  }
+}
diff --git a/examples/sparse_examples.py b/examples/sparse_examples.py
index b6d1419..35866dd 100644
--- a/examples/sparse_examples.py
+++ b/examples/sparse_examples.py
@@ -62,13 +62,46 @@ def fixtures():
     daily["datasets"][0]["coverage"] = ["complete"]*59+["incomplete"]
     daily["datasets"][0]["values"][-1] = 3
     result["unfinished_daily"] = daily
+    # Genuine distinct calls share the recorded timestamp; IDs do not order them.
+    same_time = {"id":"call-cost", "entity":{"application":"synthetic"}, "units":"dollars_per_call",
+        "timestamps":["2026-01-01T09:00:00.123Z"]*3+["2026-01-01T09:00:05.456Z"]*4+["2026-01-01T09:00:10.789Z"]*3,
+        "values":[.10,.10,.10,.09,.10,.11,.10,.80,.12,.90],
+        "event_ids":[f"call-{i:02}" for i in range(10)],
+        "data_policy":{"observation_basis":"event_index"},
+        "rules":[{"id":"illustrative-cap","kind":"acceptable_range","minimum":0,"maximum":.20}]}
+    result["identified_calls"] = request(same_time, training_size=2, calibration_size=3)
+    corrected = deepcopy(result["identified_calls"])
+    corrected["datasets"][0]["revisions"] = [{"event_id":"call-07", "timestamp":same_time["timestamps"][7],
+        "value":.85,"available_at":"2026-01-01T09:01:00Z","revision_id":"corrected-cost-1"}]
+    corrected["context"] = {"as_of":"2026-01-01T09:02:00Z"}
+    result["identified_corrected"] = corrected
+    before = deepcopy(corrected)
+    before["context"]["as_of"] = "2026-01-01T09:00:30Z"
+    result["identified_before_correction"] = before
+    tied_pattern = deepcopy(same_time)
+    tied_pattern.update(values=[.10,.10,.10,.09,.10,.11,.10]+[.12]*12,
+        timestamps=same_time["timestamps"][:7]+[same_time["timestamps"][-1]]*12,
+        event_ids=[f"call-{i:02}" for i in range(19)])
+    result["identified_pattern_ties"] = request(tied_pattern, training_size=2, calibration_size=3, point_threshold=100)
+    ordered = deepcopy(tied_pattern)
+    ordered["event_sequence"] = list(range(100,119))
+    ordered["data_policy"]["event_order"] = "trusted_sequence"
+    result["identified_pattern_sequence"] = request(ordered, training_size=2, calibration_size=3, point_threshold=100)
     return result
 
 
+def retry_fixture():
+    document = deepcopy(fixtures()["identified_calls"])
+    ds = document["datasets"][0]
+    for name in ("timestamps", "values", "event_ids"):
+        ds[name].append(ds[name][-1])
+    return document
+
+
 def generate():
     directory = Path(__file__).parent / "sparse"
     directory.mkdir(exist_ok=True)
-    for name, document in fixtures().items():
+    for name, document in {**fixtures(), "identified_retry":retry_fixture()}.items():
         (directory/f"{name}.json").write_text(json.dumps(document, indent=2)+"\n", encoding="utf-8")
 
 
@@ -94,6 +127,23 @@ def run_checks():
     assert "missing_source_value" in reasons and "zero_denominator" in reasons
     daily = next(item for item in outputs["unfinished_daily"].dataset_results if item.resolution == "completed_period")
     assert [r["complete"] for r in daily.data_quality["aggregation"]] == [False, True, False]
+    calls = outputs["identified_calls"]
+    assert {c.event_id for c in calls.cases} == {"call-07","call-09"}
+    assert not calls.dataset_results[0].anomaly_patterns
+    again = analyze(fixtures()["identified_calls"])
+    assert [c.model_dump() for c in calls.cases] == [c.model_dump() for c in again.cases]
+    assert [c.model_dump() for c in calls.cases] == [c.model_dump() for c in outputs["identified_before_correction"].cases]
+    original = next(c for c in calls.cases if c.event_id == "call-07")
+    revised = next(c for c in outputs["identified_corrected"].cases if c.event_id == "call-07")
+    assert original.id == revised.id and original.revision_id != revised.revision_id
+    assert not outputs["identified_pattern_ties"].dataset_results[0].anomaly_patterns
+    assert any(p["rule"] == "nelson_rule_2" for p in outputs["identified_pattern_sequence"].dataset_results[0].anomaly_patterns)
+    try:
+        analyze(json.loads((directory/"identified_retry.json").read_text()))
+    except ValueError as exc:
+        assert "duplicate logical" in str(exc)
+    else:
+        raise AssertionError("Retry delivery must not become another logical call")
     for name, result in outputs.items():
         print(f"PASS sparse/{name}: {result.status}; {len(result.cases)} numerical case(s)")
     return outputs
diff --git a/schemas/request-1.1.schema.json b/schemas/request-1.1.schema.json
index 84c11ec..8480b4c 100644
--- a/schemas/request-1.1.schema.json
+++ b/schemas/request-1.1.schema.json
@@ -127,6 +127,15 @@
           "title": "Observation Basis",
           "type": "string"
         },
+        "event_order": {
+          "default": "timestamp",
+          "enum": [
+            "timestamp",
+            "trusted_sequence"
+          ],
+          "title": "Event Order",
+          "type": "string"
+        },
         "reference_gaps": {
           "default": "same_phase",
           "enum": [
@@ -399,6 +408,43 @@
           },
           "title": "Revisions",
           "type": "array"
+        },
+        "event_ids": {
+          "anyOf": [
+            {
+              "items": {
+                "maxLength": 128,
+                "minLength": 1,
+                "pattern": "^\\S(?:.*\\S)?$",
+                "type": "string"
+              },
+              "type": "array"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "default": null,
+          "title": "Event Ids",
+          "uniqueItems": true
+        },
+        "event_sequence": {
+          "anyOf": [
+            {
+              "items": {
+                "maximum": 9223372036854775807,
+                "minimum": 0,
+                "type": "integer"
+              },
+              "type": "array"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "default": null,
+          "title": "Event Sequence",
+          "uniqueItems": true
         }
       },
       "required": [
@@ -1061,6 +1107,21 @@
           ],
           "title": "Coverage",
           "type": "string"
+        },
+        "event_id": {
+          "anyOf": [
+            {
+              "maxLength": 128,
+              "minLength": 1,
+              "pattern": "^\\S(?:.*\\S)?$",
+              "type": "string"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "default": null,
+          "title": "Event Id"
         }
       },
       "required": [
diff --git a/schemas/result-1.1.schema.json b/schemas/result-1.1.schema.json
index 3e6b3fd..868c16e 100644
--- a/schemas/result-1.1.schema.json
+++ b/schemas/result-1.1.schema.json
@@ -147,6 +147,18 @@
           ],
           "title": "Timestamp"
         },
+        "event_id": {
+          "anyOf": [
+            {
+              "type": "string"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "default": null,
+          "title": "Event Id"
+        },
         "classification": {
           "enum": [
             "observation",
@@ -356,6 +368,18 @@
           ],
           "title": "Event Time"
         },
+        "event_id": {
+          "anyOf": [
+            {
+              "type": "string"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "default": null,
+          "title": "Event Id"
+        },
         "detection_time": {
           "anyOf": [
             {
@@ -898,6 +922,20 @@
           },
           "title": "Source Input Indexes",
           "type": "object"
+        },
+        "source_event_ids": {
+          "additionalProperties": {
+            "type": "string"
+          },
+          "title": "Source Event Ids",
+          "type": "object"
+        },
+        "source_original_timestamps": {
+          "additionalProperties": {
+            "type": "string"
+          },
+          "title": "Source Original Timestamps",
+          "type": "object"
         }
       },
       "required": [
diff --git a/scripts/publish_release.py b/scripts/publish_release.py
index 5e2bcfd..0c63ee1 100644
--- a/scripts/publish_release.py
+++ b/scripts/publish_release.py
@@ -4,6 +4,7 @@
 from pathlib import Path
 import subprocess
 import tempfile
+import time
 
 from release import digest, verify
 
@@ -56,7 +57,14 @@ def publish(dist, tag, revision, *, finalize=False):
                 encoding="utf-8")
             gh("release", "create", tag, "--repo", REPOSITORY, "--draft", "--verify-tag",
                "--target", revision, "--title", f"Anomalyzer {tag}", "--notes-file", str(notes))
-        release = find_release(tag)
+        # GitHub's draft index can lag behind successful creation. Retry reads,
+        # never creation, and leave the resumable draft intact on timeout.
+        for attempt in range(10):
+            release = find_release(tag)
+            if release is not None:
+                break
+            if attempt < 9:
+                time.sleep(2)
         if release is None:
             raise RuntimeError("Created draft is not visible yet; rerun to resume it safely")
     expected = {p.name: p for p in dist.iterdir() if p.is_file() and p.name != ".gitignore"}
diff --git a/src/anomalyzer/_version.py b/src/anomalyzer/_version.py
index 417c5f7..8586640 100644
--- a/src/anomalyzer/_version.py
+++ b/src/anomalyzer/_version.py
@@ -1,3 +1,3 @@
 """Authoritative distribution and runtime version (read by Hatchling)."""
 
-__version__ = "0.2.1"
+__version__ = "0.3.0"
diff --git a/src/anomalyzer/agent.py b/src/anomalyzer/agent.py
index d03bc7c..3ae6655 100644
--- a/src/anomalyzer/agent.py
+++ b/src/anomalyzer/agent.py
@@ -17,7 +17,7 @@ def analyze_series(dataset: Dataset | dict | list[float | None],
             raise ValueError("context requires a Dataset rather than a bare values list")
         return analyze(dataset, settings)
     if isinstance(dataset, DatasetV11) or isinstance(dataset, dict) and any(
-            name in dataset for name in ("data_policy", "coverage", "available_at", "revisions", "rules", "completed_period_rules")):
+            name in dataset for name in ("data_policy", "coverage", "available_at", "revisions", "rules", "completed_period_rules", "event_ids", "event_sequence")):
         config = Settings.model_validate(settings or {})
         validated = DatasetV11.model_validate(dataset)
         return analyze({"schema_version": "1.1", "datasets": [validated.model_dump(mode="json")],
diff --git a/src/anomalyzer/alignment.py b/src/anomalyzer/alignment.py
index 4c791fd..a8b10ca 100644
--- a/src/anomalyzer/alignment.py
+++ b/src/anomalyzer/alignment.py
@@ -60,4 +60,11 @@ def exact_sources(ids: list[str], prepared: dict[str, PreparedDataset]):
     if any(source.times != reference for source in sources[1:]):
         raise ValueError(
             "exact relationship alignment requires identical UTC-normalized timestamps")
+    if any("event_identity_contract" in source.quality for source in sources):
+        signatures = [[(s.get("event_id"), s.get("event_sequence"))
+                       for s in source.quality.get("samples", [])] for source in sources]
+        if (not all("event_identity_contract" in source.quality for source in sources)
+                or any(signature != signatures[0] for signature in signatures[1:])
+                or len({source.quality["event_order"] for source in sources}) != 1):
+            raise ValueError("identified relationships require matching logical event IDs and order contracts")
     return sources
diff --git a/src/anomalyzer/cases.py b/src/anomalyzer/cases.py
index a558454..289fad1 100644
--- a/src/anomalyzer/cases.py
+++ b/src/anomalyzer/cases.py
@@ -43,11 +43,11 @@ def compose_cases(dataset_results: list[DatasetResultV11],
         event = assessment.timestamp if assessment.timestamp is not None else assessment.index
         entity = entities[source_id]
         entity_key = json.dumps(entity, sort_keys=True, separators=(",", ":"))
-        grouped.setdefault((event, entity_key), []).append(
+        grouped.setdefault((event, entity_key, assessment.event_id), []).append(
             (source_kind, source_id, assessment))
     cases = []
-    for (event, entity_key), items in sorted(
-            grouped.items(), key=lambda item: (str(item[0][0]), item[0][1])):
+    for (event, entity_key, event_id), items in sorted(
+            grouped.items(), key=lambda item: (str(item[0][0]), item[0][1], str(item[0][2]))):
         entity = json.loads(entity_key)
         datasets = sorted({source_id for kind, source_id, _ in items if kind == "dataset"})
         relationships = sorted({source_id for kind, source_id, _ in items if kind == "relationship"})
@@ -56,9 +56,10 @@ def compose_cases(dataset_results: list[DatasetResultV11],
                       {assessment.id for _, _, assessment in items})
         stable_material = json.dumps(
             {"event": event, "entity": entity, "datasets": datasets,
-             "relationships": relationships},
+             "relationships": relationships, **({"event_id": event_id} if event_id is not None else {})},
             sort_keys=True, separators=(",", ":"))
-        case_id = "case:" + hashlib.sha256(stable_material.encode()).hexdigest()[:16]
+        case_digest = hashlib.sha256(stable_material.encode()).hexdigest()
+        case_id = "case:" + (case_digest if event_id is not None else case_digest[:16])
         maturity_order = {
             "observation_only": 0, "early": 1, "provisional": 2, "calibrated": 3,
         }
@@ -110,13 +111,16 @@ def compose_cases(dataset_results: list[DatasetResultV11],
             "explanation": explanation,
             "action_eligible": action,
             "notification_eligible": notification,
+            **({"assessments": [assessment.model_dump(exclude={"index"})
+                for _, _, assessment in sorted(items, key=lambda entry: entry[2].id)]} if event_id is not None else {}),
         }, sort_keys=True, separators=(",", ":"))
-        revision_id = case_id + ":" + hashlib.sha256(
-            revision_material.encode()).hexdigest()[:12]
+        revision_digest = hashlib.sha256(revision_material.encode()).hexdigest()
+        revision_id = case_id + ":" + (revision_digest if event_id is not None else revision_digest[:12])
         correlation_group = sorted(
             assessment.id for _, _, assessment in items)
         cases.append(Case(
             id=case_id, revision_id=revision_id, entity=entity, event_time=event,
+            event_id=event_id,
             detection_time=max([event, *[assessment.baseline.get("known_input_availability_cutoff", event)
                                         for _, _, assessment in items]], key=str),
             source_availability_cutoff=max([event, *[assessment.baseline.get("known_input_availability_cutoff", event)
diff --git a/src/anomalyzer/contracts.py b/src/anomalyzer/contracts.py
index 4f29b02..9a4fa4b 100644
--- a/src/anomalyzer/contracts.py
+++ b/src/anomalyzer/contracts.py
@@ -243,6 +243,7 @@ class DataPolicy(Contract):
     measurement: Literal["measurement", "additive_volume"] = "measurement"
     absence_means_zero: bool = False
     observation_basis: Literal["calendar", "event_index"] = "calendar"
+    event_order: Literal["timestamp", "trusted_sequence"] = "timestamp"
     reference_gaps: Literal["same_phase", "restart"] = "same_phase"
     grid_start: str | None = None
     grid_end: str | None = None
@@ -251,6 +252,8 @@ class DataPolicy(Contract):
 
     @model_validator(mode="after")
     def meaningful(self):
+        if self.event_order == "trusted_sequence" and self.observation_basis != "event_index":
+            raise ValueError("trusted_sequence requires event_index")
         if (self.missing == "zero_for_empty" or self.absence_means_zero) and self.measurement != "additive_volume":
             raise ValueError("zero-for-empty requires measurement='additive_volume'")
         if self.absence_means_zero and self.missing != "zero_for_empty":
@@ -262,6 +265,7 @@ def meaningful(self):
 
 
 Coverage = Literal["complete", "empty", "unknown", "incomplete", "excluded", "revised"]
+EventId = Annotated[str, Field(min_length=1, max_length=128, strict=True, pattern=r"^\S(?:.*\S)?$")]
 
 
 class SampleRevision(Contract):
@@ -270,6 +274,7 @@ class SampleRevision(Contract):
     available_at: str
     revision_id: str = Field(min_length=1)
     coverage: Coverage = "revised"
+    event_id: EventId | None = None
 
     @field_validator("value", mode="before")
     @classmethod
@@ -287,10 +292,12 @@ class DatasetV11(Dataset):
     coverage: list[Coverage] | None = None
     available_at: list[str | None] | None = None
     revisions: list[SampleRevision] = Field(default_factory=list)
+    event_ids: list[EventId] | None = Field(default=None, json_schema_extra={"uniqueItems": True})
+    event_sequence: list[Annotated[int, Field(ge=0, le=9223372036854775807, strict=True)]] | None = Field(default=None, json_schema_extra={"uniqueItems": True})
 
     @model_validator(mode="after")
     def source_annotations(self):
-        for name in ("coverage", "available_at"):
+        for name in ("coverage", "available_at", "event_ids", "event_sequence"):
             annotation = getattr(self, name)
             if annotation is not None and len(annotation) != len(self.values):
                 raise ValueError(f"{name} and values must have equal length")
@@ -298,6 +305,24 @@ def source_annotations(self):
             raise ValueError("source policies/coverage/revisions require original timestamps")
         if self.data_policy and self.data_policy.observation_basis == "event_index" and self.frequency is not None:
             raise ValueError("event_index must omit frequency; event position is not elapsed cadence")
+        identified = self.event_ids is not None
+        if identified:
+            if self.timestamps is None or self.data_policy is None or self.data_policy.observation_basis != "event_index":
+                raise ValueError("event_ids require original timestamps and an explicit event_index policy")
+            if len(set(self.event_ids)) != len(self.event_ids):
+                raise ValueError("duplicate logical event IDs; deduplicate retry deliveries before analysis")
+            if any(revision.event_id is None for revision in self.revisions):
+                raise ValueError("identified event revisions require event_id")
+        elif any(revision.event_id is not None for revision in self.revisions):
+            raise ValueError("revision event_id requires dataset event_ids")
+        trusted = self.data_policy is not None and self.data_policy.event_order == "trusted_sequence"
+        if trusted and (not identified or self.event_sequence is None):
+            raise ValueError("trusted_sequence requires event_ids and aligned event_sequence")
+        if self.event_sequence is not None:
+            if not trusted:
+                raise ValueError("event_sequence requires an explicit trusted_sequence contract")
+            if len(set(self.event_sequence)) != len(self.event_sequence):
+                raise ValueError("event_sequence must contain unique ordinals")
         return self
 
     @field_validator("rules", "completed_period_rules")
@@ -445,6 +470,7 @@ class Assessment(Contract):
     target_id: str
     index: int
     timestamp: str | None
+    event_id: str | None = None
     classification: Classification
     maturity: Maturity
     basis: Basis
@@ -486,6 +512,8 @@ class LineagePoint(Contract):
     source_revision_ids: dict[str, str | None] = Field(default_factory=dict)
     source_available_at: dict[str, str | None] = Field(default_factory=dict)
     source_input_indexes: dict[str, list[int]] = Field(default_factory=dict)
+    source_event_ids: dict[str, str] = Field(default_factory=dict)
+    source_original_timestamps: dict[str, str] = Field(default_factory=dict)
 
 
 class DatasetResultV11(Contract):
@@ -533,6 +561,7 @@ class Case(Contract):
     revision_id: str
     entity: dict[str, str]
     event_time: str | int
+    event_id: str | None = None
     detection_time: str | int
     source_availability_cutoff: str | int
     contributing_dataset_ids: list[str]
diff --git a/src/anomalyzer/events.py b/src/anomalyzer/events.py
new file mode 100644
index 0000000..8ade0f2
--- /dev/null
+++ b/src/anomalyzer/events.py
@@ -0,0 +1,268 @@
+"""event-identity-v1: stable logical identity, original time, and explicit order.
+
+Timestamp/ID sorting is a reproducible layout, never an ordering assertion.
+The caller owns stable IDs, comparable populations and any trusted sequence.
+"""
+from collections import Counter
+import hashlib
+import json
+import re
+
+from .validation import timestamp
+
+
+CONTRACT = "event-identity-v1"
+
+
+def fingerprint(value):
+    return hashlib.sha256(json.dumps(value, sort_keys=True, separators=(",", ":"),
+                                     allow_nan=False).encode()).hexdigest()
+
+
+def event_token(event_id, entity):
+    return fingerprint(dict(event_id=event_id, entity=entity))
+
+
+def event_resets(points, times, policy, zone, as_of):
+    """Timestamp resets start a whole cohort; positions cannot split unordered ties."""
+    cutoff = timestamp(as_of, zone) if as_of else None
+    resolved = []
+    for point in points:
+        if isinstance(point, int):
+            if cutoff and point >= len(times):
+                continue
+            index = point
+        else:
+            date_only = bool(re.fullmatch(r"\d{4}-\d{2}-\d{2}", point))
+            normalized = timestamp(point + "T00:00:00Z" if date_only else point, zone)
+            if cutoff and normalized > cutoff:
+                continue
+            matches = [i for i, value in enumerate(times)
+                       if value[:10] == point] if date_only else [
+                           i for i, value in enumerate(times) if value == normalized.isoformat()]
+            if not matches:
+                raise ValueError("event reset must match an available timestamp")
+            if date_only and len({times[i] for i in matches}) != 1:
+                raise ValueError("date event reset must identify one timestamp cohort")
+            index = matches[0]
+        if index <= 0 or index >= len(times):
+            raise ValueError("event reset is outside available positions 1..n-1")
+        if policy.event_order != "trusted_sequence" and times[index] == times[index-1]:
+            raise ValueError("reset position cannot split an unordered timestamp cohort")
+        resolved.append(index)
+    if len(set(resolved)) != len(resolved):
+        raise ValueError("event resets resolve to duplicate positions")
+    return sorted(resolved)
+
+
+def normalize_events(ds, cfg, context, budget):
+    policy = ds.data_policy
+    cutoff = timestamp(context.as_of, ds.timezone) if context.as_of else None
+    rows = {}
+    statuses = ds.period_statuses or ["complete"] * len(ds.values)
+    for index, event_id in enumerate(ds.event_ids):
+        budget.check()
+        if event_id in rows:
+            raise ValueError("duplicate logical event IDs; deduplicate retry deliveries before analysis")
+        observed = timestamp(ds.timestamps[index], ds.timezone)
+        supplied_at = ds.available_at[index] if ds.available_at else None
+        available = timestamp(supplied_at, ds.timezone) if supplied_at else observed
+        if available < observed:
+            raise ValueError("available_at cannot precede event time")
+        value, status = ds.values[index], statuses[index]
+        coverage = ds.coverage[index] if ds.coverage else (
+            "incomplete" if status == "incomplete" else "complete" if value is not None else "unknown")
+        if status == "incomplete" and coverage != "incomplete":
+            raise ValueError("incomplete period_status conflicts with coverage")
+        rows[event_id] = dict(event_id=event_id, observed=observed,
+            original_timestamp=ds.timestamps[index], source_index=index, initial_value=value,
+            initial_available_at=available, initial_coverage=coverage,
+            event_sequence=ds.event_sequence[index] if ds.event_sequence is not None else None,
+            original_value=value, coverage=coverage, period_status=status,
+            available_at=available, availability_supplied=supplied_at is not None, revision_id=None)
+    # Validate the complete declared sequence independently of availability filtering.
+    if policy.event_order == "trusted_sequence":
+        ordered = sorted(rows.values(), key=lambda row: row["event_sequence"])
+        if any(a["observed"] > b["observed"] for a, b in zip(ordered, ordered[1:])):
+            raise ValueError("trusted event_sequence contradicts original timestamp order")
+    revision_keys, revision_ids = set(), set()
+    for revision in ds.revisions:
+        budget.check()
+        if revision.event_id not in rows:
+            raise ValueError("revision event_id must identify an existing logical event")
+        row = rows[revision.event_id]
+        observed = timestamp(revision.timestamp, ds.timezone)
+        available = timestamp(revision.available_at, ds.timezone)
+        if observed != row["observed"]:
+            raise ValueError("event revision cannot change the original event timestamp")
+        if available <= row["initial_available_at"]:
+            raise ValueError("event revision must be available strictly after the initial observation")
+        key, identity = (revision.event_id, available), (revision.event_id, revision.revision_id)
+        if key in revision_keys or identity in revision_ids:
+            raise ValueError("duplicate event revision identity or availability")
+        revision_keys.add(key)
+        revision_ids.add(identity)
+        if cutoff and available > cutoff:
+            continue
+        if available > row["available_at"]:
+            row.update(original_value=revision.value, coverage=revision.coverage,
+                       period_status="incomplete" if revision.coverage == "incomplete" else "complete",
+                       available_at=available, revision_id=revision.revision_id)
+            row["availability_supplied"] = True
+    visible = [row for row in rows.values() if not cutoff or row["available_at"] <= cutoff]
+    visible.sort(key=lambda row: (row["observed"],
+        row["event_sequence"] if policy.event_order == "trusted_sequence" else row["event_id"]))
+    times, values, rule_values, samples = [], [], [], []
+    for index, row in enumerate(visible):
+        budget.check()
+        state, original = row["coverage"], row["original_value"]
+        if state == "empty" and original not in (None, 0):
+            raise ValueError("empty coverage conflicts with nonzero observed value")
+        if state in ("complete", "revised") and original is None:
+            state = "unknown"
+        value = original if state in ("complete", "revised") else None
+        reason = None if value is not None else {
+            "empty": "empty_preserved_as_missing", "unknown": "collection_unknown",
+            "incomplete": "period_incomplete", "excluded": "explicit_model_exclusion"}[state]
+        observed = row["observed"].isoformat()
+        times.append(observed)
+        values.append(value)
+        rule_values.append(original if state == "excluded" else value)
+        selected_revision = fingerprint(dict(event_id=row["event_id"], timestamp=observed,
+            value=original, coverage=state, available_at=row["available_at"].isoformat(),
+            revision_id=row["revision_id"], event_sequence=row["event_sequence"]))
+        samples.append(dict(index=index, event_id=row["event_id"], event_sequence=row["event_sequence"],
+            original_timestamp=row["original_timestamp"], source_index=row["source_index"],
+            source_indexes=[row["source_index"]], original_value=original, initial_value=row["initial_value"],
+            original_period_status=row["period_status"], original_coverage=row["coverage"],
+            origin="revision" if row["revision_id"] else "supplied", value=value, coverage=state,
+            reason=reason, available_at=row["available_at"].isoformat() if row["availability_supplied"] else None, revision_id=row["revision_id"],
+            selected_revision=selected_revision))
+    cohorts = []
+    for i, observed in enumerate(times):
+        if not cohorts or observed != times[cohorts[-1][0]]:
+            cohorts.append([])
+        cohorts[-1].append(i)
+    ties = [group for group in cohorts if len(group) > 1]
+    ambiguous = [i for group in ties for i in group] if policy.event_order == "timestamp" else []
+    boundaries = sorted({p for i in ambiguous for p in (i, i+1) if p < len(times)})
+    quality = dict(input_count=len(ds.values), grid_count=len(times),
+        observation_count=sum(v is not None for v in rule_values),
+        reported_measurement_count=sum(v is not None for v in rule_values),
+        complete_count=sum(v is not None for v in values),
+        incomplete_count=sum(s["coverage"] == "incomplete" for s in samples),
+        incomplete_periods=[dict(index=s["index"], timestamp=times[s["index"]],
+            event_id=s["event_id"], observed=s["original_value"], period_status="incomplete")
+            for s in samples if s["coverage"] == "incomplete"],
+        missing_count=sum(v is None for v in values), duplicate_count=0,
+        regular=True, timestamp_normalization="UTC", coordinate="event_index",
+        regularity_basis="explicit_comparable_events", units=ds.units, entity=ds.entity,
+        window=[times[0], times[-1]] if times else None, data_policy=policy.model_dump(mode="json"),
+        samples=samples, rule_values=rule_values, coverage_counts=dict(Counter(s["coverage"] for s in samples)),
+        zero_filled_count=0, omitted_count=0, usable_model_count=sum(v is not None for v in values),
+        measured_value_count=sum(v is not None for v in rule_values), duplicate_sources=[],
+        latest_evidence_state="insufficient_evidence" if values and values[-1] is None else "available",
+        arrival_semantics="supplied availability, otherwise event-time snapshot; no delivery latency claim",
+        event_identity_contract=CONTRACT, event_order=policy.event_order,
+        deterministic_order="timestamp then trusted sequence" if policy.event_order == "trusted_sequence" else "timestamp then event_id (layout only)",
+        tied_timestamp_count=len(ties), tied_event_count=sum(map(len, ties)),
+        unordered_tie_positions=ambiguous, sequence_boundaries=boundaries,
+        reference_cohort_ends=[group[-1] for group in cohorts] if policy.event_order == "timestamp" else list(range(len(times))),
+        reference_restart_positions=([group[-1]+1 for group in cohorts
+            if group[-1]+1 < len(times) and any(values[i] is None for i in group)]
+            if policy.event_order == "timestamp" else [i+1 for i, v in enumerate(values[:-1]) if v is None]),
+        transformations=["Distinct logical events retained; no jitter, aggregation, interpolation or retry duplication."])
+    quality["event_reset_positions"] = event_resets(cfg.reset_points, times, policy, ds.timezone, context.as_of)
+    return times, values, quality
+
+
+def _availability_ranges(samples):
+    """Bounded O(n) storage and O(log n) queries for descriptive reference spans."""
+    size = 1 << max(0, len(samples)-1).bit_length()
+    tree = [None]*(2*size)
+    tree[size:size+len(samples)] = [s["available_at"] for s in samples]
+    def greatest(a, b):
+        return max(a, b) if a and b else a or b
+    for i in range(size-1, 0, -1):
+        tree[i] = greatest(tree[2*i], tree[2*i+1])
+    def latest(start, stop):
+        start, stop, found = start+size, stop+size, None
+        while start < stop:
+            if start%2:
+                found = greatest(found, tree[start])
+                start += 1
+            if stop%2:
+                stop -= 1
+                found = greatest(found, tree[stop])
+            start //= 2
+            stop //= 2
+        return found
+    return latest
+
+
+def annotate_assessments(assessments, quality, methods):
+    if "event_identity_contract" not in quality:
+        return
+    revisions, reference_cutoffs = {}, {}
+    latest = _availability_ranges(quality["samples"])
+    evidence_indexes = {row["id"]: row["index"] for method in methods for row in method.evidence}
+    for method in methods:
+        for segment in method.diagnostics.get("segments", []):
+            references = segment["training_indexes"] + segment["calibration_indexes"]
+            known = [quality["samples"][i]["available_at"] for i in references
+                     if quality["samples"][i]["available_at"]]
+            reference_cutoff = max(known) if known else None
+            for i in range(segment["start_position"], segment["end_position"]+1):
+                revisions[i] = segment.get("reference_revision")
+                reference_cutoffs[i] = reference_cutoff
+    for assessment in assessments:
+        sample = quality["samples"][assessment.index]
+        assessment.event_id = sample["event_id"]
+        assessment.id = assessment.id.replace(f":assessment:{assessment.index}:",
+            f":assessment:event:{event_token(sample['event_id'], quality['entity'])}:", 1)
+        assessment.baseline["event_identity"] = dict(contract=CONTRACT,
+            event_id=sample["event_id"], original_timestamp=sample["original_timestamp"],
+            event_sequence=sample["event_sequence"], revision_id=sample["revision_id"],
+            selected_revision=sample["selected_revision"], event_order=quality["event_order"])
+        statistical = assessment.baseline["kind"] in ("seasonal_residual", "location_shift_pattern")
+        if statistical:
+            assessment.baseline["reference_revision"] = revisions.get(assessment.index)
+        # A same-time sibling is not an input to this point's reference or rule.
+        # Replace the generic conservative prefix cutoff with actual support.
+        available = [sample["available_at"]]
+        if statistical:
+            available.append(reference_cutoffs.get(assessment.index))
+            if assessment.baseline["kind"] == "location_shift_pattern":
+                available += [quality["samples"][evidence_indexes[ref]]["available_at"]
+                    for ref in assessment.evidence_refs if ref in evidence_indexes]
+        elif assessment.baseline["kind"] == "explicit_rules":
+            if any(rule["kind"] != "acceptable_range" for rule in assessment.baseline["violations"]):
+                available.append(quality["samples"][assessment.index-1]["available_at"])
+        else:
+            available.append(latest(max(0, assessment.index-assessment.comparison_samples), assessment.index))
+        available = [value for value in available if value]
+        assessment.baseline.pop("known_input_availability_cutoff", None)
+        if available:
+            assessment.baseline["known_input_availability_cutoff"] = max([assessment.timestamp, *available])
+        assessment.assumptions = [text for text in assessment.assumptions
+            if text != "Input order is chronological after preparation."
+            and not text.startswith("Availability cutoff is conservative across known prior inputs")]
+        assessment.assumptions.append("Event IDs identify observations, not independent confirmations; ID tie order establishes no chronology or elapsed rate.")
+
+
+def identify_evidence(methods, observations, patterns, quality):
+    """Stable IDs and bounded lineage live beside the original index/time contract."""
+    mapping = {}
+    samples = quality["samples"]
+    for method in methods:
+        for item in method.evidence:
+            mapping[item["id"]] = f"{method.id}:event:{event_token(samples[item['index']]['event_id'], quality['entity'])}"
+            item["id"] = mapping[item["id"]]
+    for observation in observations:
+        ids = [samples[i]["event_id"] for i in observation["triggering_samples"]]
+        observation["id"] = "event-episode:" + fingerprint([quality["entity"], ids])
+        observation["evidence_refs"] = [mapping[ref] for ref in observation["evidence_refs"]]
+    for pattern in patterns:
+        pattern["id"] = "event-pattern:" + fingerprint([
+            quality["entity"], pattern["rule"], pattern["direction"], samples[pattern["detection_index"]]["event_id"]])
+        pattern["evidence_refs"] = [mapping[ref] for ref in pattern["evidence_refs"]]
diff --git a/src/anomalyzer/mcp.py b/src/anomalyzer/mcp.py
index 58f0238..34f1457 100644
--- a/src/anomalyzer/mcp.py
+++ b/src/anomalyzer/mcp.py
@@ -56,10 +56,24 @@ class GetEvidenceArguments(ListFindingsArguments):
     "previously reported evidence using caller-held state. Policy eligibility "
     "does not check detection readiness and is not an action recommendation. "
     "The server does not schedule runs, deliver alerts, or execute actions. "
+    "Identified same-time events remain separate points; IDs are not temporal "
+    "order or independent confirmations. Inspect source_samples and event_order. "
     "Do not infer cause, business impact, or normality from absent findings."
 )
 
 
+def _input_schema(model):
+    """Drop display titles from wire schemas, retaining all validation constraints."""
+    def compact(value):
+        if isinstance(value, dict):
+            return {key: compact(item) for key, item in value.items()
+                    if key != "title" or not isinstance(item, str)}
+        if isinstance(value, list):
+            return [compact(item) for item in value]
+        return value
+    return compact(model.model_json_schema())
+
+
 TOOLS = [
     {
         "name": "analyze",
@@ -68,7 +82,7 @@ class GetEvidenceArguments(ListFindingsArguments):
             "assessment, readiness and data-quality checks, a preview of findings, "
             "and a run_id for drilldown. The full result stays in this MCP session. "
             "No finding establishes cause, incident status, or action authority."),
-        "inputSchema": AnalyzeRequestArguments.model_json_schema(),
+        "inputSchema": _input_schema(AnalyzeRequestArguments),
         "outputSchema": {
             "type": "object",
             "required": ["run_id", "schema_version", "status", "headline",
@@ -94,7 +108,7 @@ class GetEvidenceArguments(ListFindingsArguments):
             "Page through cases, early candidates, point episodes, residual patterns, and data-quality "
             "issues from a prior analyze call. These types can overlap and are not "
             "a count of separate incidents. The run_id lasts only for this session."),
-        "inputSchema": ListFindingsArguments.model_json_schema(),
+        "inputSchema": _input_schema(ListFindingsArguments),
         "outputSchema": {
             "type": "object", "additionalProperties": True,
             "required": ["run_id", "total", "findings", "next_cursor"],
@@ -113,7 +127,7 @@ class GetEvidenceArguments(ListFindingsArguments):
             "Inspect one finding by run_id and finding_id. Returns the numerical "
             "record, relevant assessments and lineage, readiness, limitations, "
             "and a page of source samples. This does not establish cause or impact."),
-        "inputSchema": GetEvidenceArguments.model_json_schema(),
+        "inputSchema": _input_schema(GetEvidenceArguments),
         "outputSchema": {
             "type": "object", "additionalProperties": True,
             "required": ["run_id", "finding", "record", "sample_count",
diff --git a/src/anomalyzer/mcp_views.py b/src/anomalyzer/mcp_views.py
index 627a682..d5c2e77 100644
--- a/src/anomalyzer/mcp_views.py
+++ b/src/anomalyzer/mcp_views.py
@@ -73,6 +73,8 @@ def _quality(item):
             "zero_filled_count": source["zero_filled_count"], "observation_basis": source["coordinate"]}
            if "data_policy" in source else {}),
         **({"resolution": item["resolution"]} if item.get("resolution") else {}),
+        **({key: source[key] for key in ("event_identity_contract", "event_order", "tied_timestamp_count", "tied_event_count")}
+           if "event_identity_contract" in source else {}),
     }
 
 
@@ -147,6 +149,7 @@ def build_findings(result):
                 "id": case["id"], "kind": "case", "target_id": target,
                 **({"resolution": case["resolution"]} if case.get("resolution") else {}),
                 "entity": case["entity"], "event_time": case["event_time"],
+                **({"event_id": case["event_id"]} if case.get("event_id") is not None else {}),
                 "headline": _case_headline(case, matches, definitions),
                 "maturity": case["maturity"],
                 "severity": case["severity"],
@@ -371,6 +374,12 @@ def get_evidence(result, finding_id, cursor=0, limit=50):
         "run_id": result["run_id"], "finding": finding,
         "record": record, "assessments": assessments,
         "lineage": lineage, "readiness": readiness,
+        "source_samples": [dict(target_id=target, **point)
+            for _, target, target_item in related
+            for point in target_item["data_quality"].get("samples", [])
+            if point.get("event_id") is not None and point["index"] in (
+                {sample["index"] for sample in selected_samples} |
+                {assessment["index"] for assessment in assessments if assessment["target_id"] == target})][:limit],
         "sample_count": len(samples), "samples": selected_samples,
         "next_cursor": end if end < len(samples) else None,
         "note": "Evidence establishes numerical criteria only; source context, cause, impact, and action authority require external review.",
diff --git a/src/anomalyzer/orchestrator.py b/src/anomalyzer/orchestrator.py
index 4a7b268..e1021ff 100644
--- a/src/anomalyzer/orchestrator.py
+++ b/src/anomalyzer/orchestrator.py
@@ -21,6 +21,7 @@
 from .relationships import derive, _source_ids
 from .validation import resolve_reset_points
 from .sparse import analyze_sparse, active_resets
+from .events import annotate_assessments
 
 
 def annotate_availability(assessments, times, samples):
@@ -138,7 +139,9 @@ def analyze_relationships(request: RequestV11 | dict, *, _budget=None) -> Result
                 source.times[:source.quality.get("grid_count", source.quality["observation_count"])],
                 source.quality.get("rule_values", source.values)[:source.quality.get("grid_count", source.quality["observation_count"])], dataset.rules,
                 request.policy,
-                reset_positions=active_resets(request.config.reset_points, source.times[:source.quality.get("grid_count", source.quality["observation_count"])], dataset.timezone, request.context.as_of),
+                reset_positions=(source.quality["event_reset_positions"] + source.quality["sequence_boundaries"]
+                    if "event_identity_contract" in source.quality else
+                    active_resets(request.config.reset_points, source.times[:source.quality.get("grid_count", source.quality["observation_count"])], dataset.timezone, request.context.as_of)),
                 change_max_gap_seconds=dataset.data_policy.change_max_gap_seconds if dataset.data_policy else None,
                 excluded_positions=[s["index"] for s in source.quality.get("samples", []) if s["coverage"] == "excluded"])
             assessments.extend(calibrated_assessments(
@@ -148,6 +151,7 @@ def analyze_relationships(request: RequestV11 | dict, *, _budget=None) -> Result
                     dataset.id, result.methods, result.anomaly_patterns,
                     request.policy))
             annotate_availability(assessments, source.times, source.quality.get("samples", []))
+            annotate_assessments(assessments, source.quality, result.methods)
             dataset_results.append(DatasetResultV11(
                 dataset_id=dataset.id, status=result.status,
                 data_quality=result.data_quality, methods=result.methods,
@@ -173,11 +177,22 @@ def analyze_relationships(request: RequestV11 | dict, *, _budget=None) -> Result
                 analysis_view, request.config, request.context, budget=budget)
             _shift_result_indexes(result, analysis_offset)
             _namespace_result(result, f"relationship:{relationship.id}")
+            if "event_identity_contract" in result.data_quality:
+                from .events import fingerprint as event_fingerprint
+                for method in result.methods:
+                    for segment in method.diagnostics.get("segments", []):
+                        if segment.get("reference_revision"):
+                            segment["reference_revision"] = event_fingerprint([
+                                segment["reference_revision"],
+                                [lineage[i].source_revision_ids for i in
+                                 segment["training_indexes"] + segment["calibration_indexes"]]])
             quality["analysis_offset"] = analysis_offset
             assessments = descriptive_assessments(
                 relationship.id, derived.timestamps or [], derived.values,
                 relationship.rules, request.policy, relationship=True,
-                reset_positions=active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of),
+                reset_positions=(result.data_quality["event_reset_positions"] + result.data_quality["sequence_boundaries"]
+                    if "event_identity_contract" in result.data_quality else
+                    active_resets(request.config.reset_points, derived.timestamps or [], derived.timezone, request.context.as_of)),
                 change_max_gap_seconds=getattr(getattr(derived, "data_policy", None), "change_max_gap_seconds", None),
                 excluded_positions=[i for i, state in enumerate(getattr(derived, "coverage", None) or []) if state == "excluded"])
             assessments.extend(calibrated_assessments(
@@ -189,6 +204,13 @@ def analyze_relationships(request: RequestV11 | dict, *, _budget=None) -> Result
                     request.policy, relationship=True))
             annotate_availability(assessments, [p.timestamp for p in lineage],
                 [dict(value=p.value, available_at=p.latest_source_cutoff if any(p.source_available_at.values()) else None) for p in lineage])
+            annotate_assessments(assessments, result.data_quality, result.methods)
+            for assessment in assessments:
+                if assessment.event_id is not None:
+                    assessment.baseline["source_revision_ids"] = lineage[assessment.index].source_revision_ids
+            if "event_identity_contract" in result.data_quality:
+                quality.update({key: result.data_quality[key] for key in (
+                    "samples", "event_identity_contract", "event_order", "tied_timestamp_count", "tied_event_count")})
             quality["model_eligibility"] = {key: result.data_quality.get(key)
                 for key in ("coordinate", "usable_model_count", "coverage_counts")}
             if quality["available_count"] == 0 or quality["latest_evidence_state"] == "insufficient_evidence":
diff --git a/src/anomalyzer/relationships.py b/src/anomalyzer/relationships.py
index 3403126..4f54c81 100644
--- a/src/anomalyzer/relationships.py
+++ b/src/anomalyzer/relationships.py
@@ -32,6 +32,10 @@ def derive(relationship: Relationship,
            prepared: dict[str, PreparedDataset]) -> tuple[Dataset, list[LineagePoint], dict]:
     ids = _source_ids(relationship)
     sources = exact_sources(ids, prepared)
+    identified = "event_identity_contract" in sources[0].quality
+    if (identified and relationship.kind == "lagged_response"
+            and sources[0].quality["unordered_tie_positions"]):
+        raise ValueError("lagged_response cannot order tied events without trusted_sequence")
     # Exclusion from a model does not erase a genuine measured value.
     sources = [replace(source, values=source.quality.get("rule_values", source.values)) for source in sources]
     by_id = {source.dataset.id: source for source in sources}
@@ -177,6 +181,9 @@ def derive(relationship: Relationship,
             source.quality["samples"][i]["available_at"] if "samples" in source.quality else None for source, i in used}
         lineage[-1].source_input_indexes = {source.dataset.id:
             source.quality["samples"][i]["source_indexes"] if "samples" in source.quality else [source.source_indexes[i]] for source, i in used}
+        if identified:
+            lineage[-1].source_event_ids = {source.dataset.id: source.quality["samples"][i]["event_id"] for source, i in used}
+            lineage[-1].source_original_timestamps = {source.dataset.id: source.quality["samples"][i]["original_timestamp"] for source, i in used}
         known = [lineage[-1].source_available_at[source.dataset.id] for source, i in used
                  if source.values[i] is not None and lineage[-1].source_available_at[source.dataset.id]]
         lineage[-1].latest_source_cutoff = max([cutoff, *known])
@@ -188,10 +195,14 @@ def derive(relationship: Relationship,
         derived = DatasetV11(id=relationship.id, timestamps=times, values=values,
             frequency=first.frequency, units=relationship.units, entity=first.entity, timezone=first.timezone,
             data_policy=DataPolicy(observation_basis=basis,
+                event_order=policies[0].event_order if identified else "timestamp",
                 reference_gaps="restart" if any(p.reference_gaps == "restart" for p in policies) else "same_phase",
                 minimum_phase_samples=max(p.minimum_phase_samples for p in policies),
                 change_max_gap_seconds=min(separation) if separation else None),
-            coverage=["excluded" if i in excluded_indexes else "complete" if v is not None else "unknown" for i, v in enumerate(values)])
+            coverage=["excluded" if i in excluded_indexes else "complete" if v is not None else "unknown" for i, v in enumerate(values)],
+            event_ids=[s["event_id"] for s in sources[0].quality["samples"]] if identified else None,
+            available_at=[p.latest_source_cutoff if any(p.source_available_at.values()) else None for p in lineage] if identified else None,
+            event_sequence=[s["event_sequence"] for s in sources[0].quality["samples"]] if identified and policies[0].event_order == "trusted_sequence" else None)
     else:
         derived = Dataset(id=relationship.id, timestamps=times, values=values,
             frequency=first.frequency, units=relationship.units, entity=first.entity, timezone=first.timezone)
diff --git a/src/anomalyzer/sparse.py b/src/anomalyzer/sparse.py
index 47c4abf..638747b 100644
--- a/src/anomalyzer/sparse.py
+++ b/src/anomalyzer/sparse.py
@@ -43,6 +43,9 @@ def normalize(ds, cfg, context, budget=None):
     budget = budget or RuntimeBudget(cfg.max_runtime_seconds)
     policy = ds.data_policy or DataPolicy()
     context = Context.model_validate(context or {})
+    if ds.event_ids is not None:
+        from .events import normalize_events
+        return normalize_events(ds, cfg, context, budget)
     cutoff = timestamp(context.as_of, ds.timezone) if context.as_of else None
     rows = {}
     statuses = ds.period_statuses or ["complete"] * len(ds.values)
@@ -188,13 +191,15 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
     context = Context.model_validate(context or {})
     times, values, quality = prepared or normalize(ds, cfg, context, budget)
     policy = ds.data_policy or DataPolicy()
-    resets = active_resets(cfg.reset_points, times, ds.timezone, context.as_of)
+    identified = "event_identity_contract" in quality
+    resets = quality["event_reset_positions"] if identified else active_resets(cfg.reset_points, times, ds.timezone, context.as_of)
+    cohort_ends = set(quality.get("reference_cohort_ends", range(len(values))))
     if policy.observation_basis == "calendar" and (cfg.recipe == "adaptive-seasonal-v1" or cfg.trend not in ("none", "auto")):
         return analyze_contiguous(ds, cfg, context, budget, (times, values, quality), resets)
     method = MethodResult(id="sparse_seasonal", status="insufficient_history",
         applicability="eligible observations on the original grid; frozen training phase reference",
         parameters={**cfg.model_dump(), "resolved_reset_positions": resets,
-                    "profile_revision": "sparse-frozen-phase-v1",
+                    "profile_revision": "identified-events-v1" if identified else "sparse-frozen-phase-v1",
                     "observation_basis": policy.observation_basis, "reference_updates": "frozen_until_explicit_reset"},
         diagnostics={"segments": [], "unavailable": [], "pattern_boundaries": resets.copy()})
     unsupported = ("event_index requires season_length=1, trend=none/auto and a single-resolution seasonal recipe"
@@ -204,7 +209,8 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
         if cfg.recipe != "seasonal-residual-v1" or cfg.trend not in ("none", "auto") else None)
     patterns, observations = [], []
     boundaries = sorted(set([0, *resets, len(values)] + (
-        [i + 1 for i, value in enumerate(values[:-1]) if value is None]
+        (quality["reference_restart_positions"] if identified else
+         [i + 1 for i, value in enumerate(values[:-1]) if value is None])
         if policy.reference_gaps == "restart" else [])))
     try:
         if unsupported:
@@ -213,6 +219,8 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
             for segment_number, (start, stop) in enumerate(zip(boundaries, boundaries[1:])):
                 budget.check()
                 training, calibration = [], []
+                training_complete = calibration_complete = False
+                cohort_calibration_count = 0
                 phase_values, phase_indexes, phase_expected = {}, {}, {}
                 center = scale = None
                 evidence = []
@@ -222,14 +230,24 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
                     status="insufficient_history", trend="none (explicit sparse frozen reference)")
                 for index in range(start, stop):
                     budget.check()
+                    if index == start or index-1 in cohort_ends:
+                        cohort_calibration_count = len(calibration)
+                        if len(calibration) >= cfg.calibration_size and not calibration_complete:
+                            calibration_complete = True
+                            if identified:
+                                from .events import fingerprint
+                                segment["reference_revision"] = fingerprint([
+                                    quality["samples"][i]["selected_revision"]
+                                    for i in training + segment["calibration_indexes"]])
                     value = values[index]
                     phase = (index - start) % cfg.season_length
-                    if len(training) < cfg.training_size:
+                    if not training_complete:
                         if value is not None:
                             training.append(index)
                             phase_values.setdefault(phase, []).append(value)
                             phase_indexes.setdefault(phase, []).append(index)
-                        if len(training) == cfg.training_size:
+                        if len(training) >= cfg.training_size and index in cohort_ends:
+                            training_complete = True
                             phase_expected = {p: (safe_median(v) if cfg.outlier_handling == "robust" else statistics.mean(v)) for p, v in phase_values.items()
                                               if len(v) >= policy.minimum_phase_samples}
                             segment["training_indexes"] = training.copy()
@@ -250,13 +268,22 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
                         continue
                     maturity, z, triggers = "reference_only", None, []
                     count = len(calibration)
-                    if count < cfg.calibration_size:
-                        if count >= 3:
+                    if identified and not calibration_complete:
+                        count = cohort_calibration_count
+                    if not calibration_complete:
+                        if count >= 3 and not identified:
                             raw = [item[1] for item in calibration]
                             z = (residual - statistics.mean(raw)) / max(statistics.stdev(raw), cfg.scale_floor)
                             maturity = "provisional"
                         calibration.append((index, residual))
                         segment["calibration_indexes"].append(index)
+                        if len(calibration) >= cfg.calibration_size and index in cohort_ends:
+                            calibration_complete = True
+                            if identified:
+                                from .events import fingerprint
+                                segment["reference_revision"] = fingerprint([
+                                    quality["samples"][i]["selected_revision"]
+                                    for i in training + segment["calibration_indexes"]])
                     else:
                         if center is None:
                             raw = [item[1] for item in calibration]
@@ -291,12 +318,20 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
                 method.diagnostics["segments"].append(segment)
                 method.evidence.extend(evidence)
                 temporary = method.model_copy(update={"evidence": evidence})
+                if identified:
+                    # Unordered tied calls supply points, never a sequence detector.
+                    blocked = set(quality["unordered_tie_positions"])
+                    temporary = method.model_copy(update={"evidence": [e for e in evidence if e["index"] not in blocked]})
                 # Separate calls prohibit patterns/episodes crossing explicit regime boundaries.
                 for pattern in anomaly_patterns([temporary], ds.id, cusum_k=cfg.cusum_k,
                         cusum_h=cfg.cusum_h, moving_range_threshold=cfg.moving_range_threshold):
+                    if identified and pattern["sample_count"] < 2:
+                        continue
                     pattern["id"] = f"segment:{segment_number}:{pattern['id']}"
                     patterns.append(pattern)
-                for observation in episodes([temporary], ds.id):
+                episode_rows = (o for e in evidence for o in episodes(
+                    [method.model_copy(update={"evidence": [e]})], ds.id)) if identified else episodes([temporary], ds.id)
+                for observation in episode_rows:
                     observation["id"] = f"segment:{segment_number}:{observation['id']}"
                     observations.append(observation)
             if any(row["signal_maturity"] == "calibrated" for row in method.evidence):
@@ -305,6 +340,9 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
         method.status, method.error = "budget_exceeded", str(exc)
     except (ArithmeticError, ValueError) as exc:
         method.status, method.error = "failed", f"{type(exc).__name__}: {exc}"
+    if identified:
+        from .events import identify_evidence
+        identify_evidence([method], observations, patterns, quality)
     method.diagnostics["detection_readiness"] = dict(
         status="caution" if method.status == "completed" else "not_assessed",
         reasons=["Sparse frozen phase medians use only eligible training observations; gaps do not count as measurements.",
@@ -314,6 +352,10 @@ def analyze_sparse(ds, cfg, context, budget=None, prepared=None):
     method.limitations = ["Training and calibration origins remain fixed within each explicit regime; no evaluated episode updates the reference.",
         "Sparse calibration counts usable residuals, not elapsed grid positions. Phase support and readiness remain separate from business severity.",
         "No probability, health verdict, abuse accuracy, operational permission or delivery-time performance is established."]
+    if identified:
+        method.limitations.append("event-identity-v1: whole timestamp cohorts share prior-time references unless trusted_sequence is explicit. Unordered tied calls cannot supply change or sequence-pattern continuity. Distinct events are not independent confirmations.")
+        method.diagnostics["detection_readiness"]["reasons"].append(
+            "Identified sequence patterns need at least two ordered evaluation points. Tied timestamp cohorts are excluded unless trusted_sequence is declared.")
     resolved = dict(datasets=[ds.model_dump(mode="json")], config=cfg.model_dump(mode="json"),
                     context=context.model_dump(mode="json"))
     method.runtime_seconds = time.perf_counter() - started
diff --git a/tests/test_events.py b/tests/test_events.py
new file mode 100644
index 0000000..a7ea6d4
--- /dev/null
+++ b/tests/test_events.py
@@ -0,0 +1,380 @@
+"""Logical event identity, honest tie semantics and deterministic consumer replay."""
+from copy import deepcopy
+from datetime import datetime, timedelta, timezone
+import json
+from pathlib import Path
+import subprocess
+import sys
+
+import jsonschema
+import pytest
+
+from anomalyzer import analyze, analyze_series, get_case, replay_policy
+from anomalyzer.contracts import RequestV11, ResultV11
+from anomalyzer.mcp import handle, TOOLS
+from anomalyzer.mcp_views import build_findings, get_evidence, summarize
+
+
+def request(values=None, cohorts=None, trusted=False):
+    values = [10, 10, 9, 10, 11, 30, 40] if values is None else values
+    cohorts = [0, 0, 1, 1, 1, 2, 2] if cohorts is None else cohorts
+    start = datetime(2026, 1, 1, tzinfo=timezone.utc)
+    ds = dict(id="cost", entity={"application": "synthetic"}, units="dollars_per_call",
+        timestamps=[(start+timedelta(hours=i)).isoformat() for i in cohorts], values=values,
+        event_ids=[f"call-{i:04}" for i in range(len(values))],
+        data_policy={"observation_basis": "event_index"},
+        rules=[dict(id="cap", kind="acceptable_range", minimum=0, maximum=20),
+               dict(id="jump", kind="maximum_absolute_change", threshold=5)])
+    if trusted:
+        ds["event_sequence"] = list(range(100, 100+len(values)))
+        ds["data_policy"]["event_order"] = "trusted_sequence"
+    return dict(schema_version="1.1", datasets=[ds],
+                config=dict(training_size=2, calibration_size=3, trend="none", season_length=1))
+
+
+def rows(result):
+    return [item for method in result.dataset_results[0].methods for item in method.evidence]
+
+
+def semantic(value):
+    if isinstance(value, dict):
+        return {k: semantic(v) for k, v in value.items() if k not in ("run_id", "runtime_seconds", "source_index", "source_indexes")}
+    if isinstance(value, list):
+        return [semantic(v) for v in value]
+    return value
+
+
+def test_distinct_same_time_calls_keep_values_references_and_cases():
+    result = analyze(request())
+    item = result.dataset_results[0]
+    assert item.status == "completed"
+    assert item.data_quality["event_identity_contract"] == "event-identity-v1"
+    assert [s["original_value"] for s in item.data_quality["samples"]] == [10,10,9,10,11,30,40]
+    assert item.data_quality["observation_count"] == 7
+    points = [e for e in rows(result) if e["signal_maturity"] == "calibrated"]
+    assert len(points) == 2 and {e["expected"] for e in points} == {10}
+    assert all(e["training_cutoff"] < e["timestamp"] for e in points)
+    assert {e["observed"] for e in points} == {30,40}
+    assert len({e["id"] for e in points}) == 2
+    assert len(result.cases) == 2 and len({c.id for c in result.cases}) == 2
+    assert {c.event_id for c in result.cases} == {"call-0005", "call-0006"}
+    assert len({c.event_time for c in result.cases}) == 1
+    assert all(not c.action_eligible for c in result.cases)
+    assert not item.anomaly_patterns
+    assert all(len(o["triggering_samples"]) == 1 for o in item.observations)
+    for case in result.cases:
+        assert get_case(result, case.id) == case
+        detail = get_evidence(result.model_dump(mode="json"), case.id)
+        assert {s["event_id"] for s in detail["source_samples"]} == {case.event_id}
+
+
+@pytest.mark.parametrize("mutation", [
+    lambda d: d["event_ids"].__setitem__(1, d["event_ids"][0]),
+    lambda d: d["event_ids"].pop(),
+    lambda d: d["event_ids"].__setitem__(0, ""),
+    lambda d: d["event_ids"].__setitem__(0, " padded "),
+    lambda d: d["event_ids"].__setitem__(0, 4),
+    lambda d: d["event_ids"].__setitem__(0, "x"*129),
+    lambda d: d.update(frequency="1h"),
+    lambda d: d["data_policy"].update(observation_basis="calendar"),
+    lambda d: d.update(event_sequence=list(range(7))),
+    lambda d: d["data_policy"].update(event_order="trusted_sequence"),
+])
+def test_invalid_identity_contract_is_rejected(mutation):
+    body = request()
+    mutation(body["datasets"][0])
+    with pytest.raises(ValueError):
+        RequestV11.model_validate(body)
+
+
+def test_whole_cohorts_cannot_train_or_calibrate_one_another():
+    body = request([10,10,10,9,10,11,9,10,11,40], [0,0,0,1,1,1,1,2,2,3])
+    item = analyze(body).dataset_results[0]
+    segment = item.methods[0].diagnostics["segments"][0]
+    assert segment["training_indexes"] == [0,1,2]  # target 2 cannot split the cohort
+    assert segment["calibration_indexes"] == [3,4,5,6]  # target 3 cannot split it either
+    assert all(e["signal_maturity"] == "reference_only" for e in item.methods[0].evidence if e["index"] < 7)
+    assert all(e["signal_maturity"] == "calibrated" for e in item.methods[0].evidence if e["index"] >= 7)
+    assert len({e["calibration_samples"] for e in item.methods[0].evidence if e["timestamp"] == body["datasets"][0]["timestamps"][3]}) == 1
+
+
+def test_missing_last_cohort_member_does_not_pull_next_timestamp_into_calibration():
+    body = request([10,10,9,10,11,None,40], [0,0,1,1,1,1,2])
+    result = analyze(body)
+    assert rows(result)[-1]["signal_maturity"] == "calibrated"
+    assert rows(result)[-1]["triggers"] == ["point"]
+
+
+def test_unordered_ties_break_change_and_pattern_continuity_but_keep_ranges():
+    body = request([10,10,9,10,11]+[40]*12+[50], [0,0,1,1,1]+[2]*12+[3])
+    result = analyze(body)
+    item = result.dataset_results[0]
+    assert not item.anomaly_patterns
+    for assessment in item.assessments:
+        if assessment.id.endswith(":early") and assessment.index >= 5:
+            assert assessment.absolute_change is None
+            assert {v["id"] for v in assessment.baseline["violations"]} == {"cap"}
+            assert assessment.baseline["unavailable_rules"] == [{"id":"jump", "reason":"no_comparable_prior"}]
+    assert len(result.cases) == 13
+
+
+def test_trusted_sequence_is_explicit_and_enables_order_without_changing_time():
+    body = request([10,10,9,10,11]+[12]*12, [0,0,1,1,1]+[2]*12, trusted=True)
+    body["config"]["point_threshold"] = 100
+    result = analyze(body)
+    item = result.dataset_results[0]
+    assert any(p["rule"] == "nelson_rule_2" for p in item.anomaly_patterns)
+    assert {s["original_timestamp"] for s in item.data_quality["samples"][5:]} == {body["datasets"][0]["timestamps"][5]}
+    assert item.data_quality["event_order"] == "trusted_sequence"
+    assert not item.data_quality["unordered_tie_positions"]
+
+
+@pytest.mark.parametrize("sequence", [[1]*7, [False,1,2,3,4,5,6], [6,5,4,3,2,1,0]])
+def test_bad_trusted_sequence_is_rejected_or_fails_visibly(sequence):
+    body = request(trusted=True)
+    body["datasets"][0]["event_sequence"] = sequence
+    try:
+        result = analyze(body)
+    except ValueError:
+        return
+    assert result.dataset_results[0].status == "failed"
+    assert "contradicts" in result.dataset_results[0].error
+
+
+def test_reordering_deliveries_does_not_change_evidence_or_case_identity():
+    body = request()
+    original = analyze(body)
+    reordered = deepcopy(body)
+    ds = reordered["datasets"][0]
+    for name in ("timestamps", "values", "event_ids"):
+        ds[name] = list(reversed(ds[name]))
+    again = analyze(reordered)
+    assert rows(original) == rows(again)
+    assert [c.model_dump() for c in original.cases] == [c.model_dump() for c in again.cases]
+    assert [a.model_dump() for a in original.dataset_results[0].assessments] == [a.model_dump() for a in again.dataset_results[0].assessments]
+    assert semantic(original.dataset_results[0].data_quality) == semantic(again.dataset_results[0].data_quality)
+
+
+def test_repeat_replay_and_duplicate_delivery_do_not_create_new_logical_events():
+    body = request()
+    a, b = analyze(body), analyze(body)
+    assert [c.model_dump() for c in a.cases] == [c.model_dump() for c in b.cases]
+    replayed = replay_policy(a, {"allow_explicit_rules":True})
+    assert [c.id for c in replayed.cases] == [c.id for c in a.cases]
+    assert all(c.action_eligible for c in replayed.cases)
+    ds = body["datasets"][0]
+    for name in ("timestamps", "values", "event_ids"):
+        ds[name].append(ds[name][-1])
+    with pytest.raises(ValueError, match="duplicate logical"):
+        analyze(body)
+
+
+def test_late_same_time_event_does_not_change_an_earlier_asof_snapshot():
+    body = request()
+    body["context"] = {"as_of":"2026-01-01T03:00:00Z"}
+    original = analyze(body)
+    ds = body["datasets"][0]
+    ds["available_at"] = [None]*7+["2026-01-01T04:00:00Z"]
+    ds["event_ids"].append("a-new-layout-first")
+    ds["values"].append(35)
+    ds["timestamps"].append(ds["timestamps"][-1])
+    earlier = analyze(body)
+    assert rows(original) == rows(earlier)
+    assert [c.model_dump() for c in original.cases] == [c.model_dump() for c in earlier.cases]
+    body["context"]["as_of"] = "2026-01-01T05:00:00Z"
+    later = analyze(body)
+    assert len(later.cases) == 3
+    old = {c.event_id:c for c in original.cases}
+    for case in later.cases:
+        if case.event_id in old:
+            assert case.model_dump() == old[case.event_id].model_dump()
+    assert {e["id"] for e in rows(original)} <= {e["id"] for e in rows(later)}
+
+
+def test_prefix_extension_at_a_later_timestamp_preserves_prior_evidence():
+    body = request()
+    original = analyze(body)
+    ds = body["datasets"][0]
+    ds["event_ids"].append("call-last")
+    ds["values"].append(10)
+    ds["timestamps"].append("2026-01-01T03:00:00Z")
+    later = analyze(body)
+    assert rows(original) == rows(later)[:-1]
+    assert [c.model_dump() for c in original.cases] == [c.model_dump() for c in later.cases]
+
+
+def test_identified_correction_changes_only_the_addressed_event_and_case_revision():
+    body = request()
+    body["context"] = {"as_of":"2026-01-01T03:00:00Z"}
+    original = analyze(body)
+    ds = body["datasets"][0]
+    ds["revisions"] = [dict(event_id="call-0005", timestamp=ds["timestamps"][5], value=35,
+        available_at="2026-01-01T04:00:00Z", revision_id="corrected-a")]
+    earlier = analyze(body)
+    assert [c.model_dump() for c in original.cases] == [c.model_dump() for c in earlier.cases]
+    body["context"]["as_of"] = "2026-01-01T05:00:00Z"
+    later = analyze(body)
+    a = next(c for c in original.cases if c.event_id == "call-0005")
+    b = next(c for c in later.cases if c.event_id == "call-0005")
+    assert a.id == b.id and a.revision_id != b.revision_id
+    sample = later.dataset_results[0].data_quality["samples"][5]
+    assert sample["initial_value"] == 30 and sample["original_value"] == 35
+    assert sample["revision_id"] == "corrected-a"
+    assert later.dataset_results[0].data_quality["observation_count"] == 7
+    assert next(c for c in later.cases if c.event_id == "call-0006").model_dump() == next(c for c in original.cases if c.event_id == "call-0006").model_dump()
+
+
+@pytest.mark.parametrize("override,reason", [
+    ({"event_id":None}, "require event_id"),
+    ({"event_id":"unknown"}, "existing logical"),
+    ({"timestamp":"2026-01-01T01:00:00Z"}, "original event timestamp"),
+    ({"available_at":"2026-01-01T02:00:00Z"}, "strictly after"),
+])
+def test_ambiguous_corrections_are_rejected(override, reason):
+    body = request()
+    ds = body["datasets"][0]
+    ds["revisions"] = [{**dict(event_id="call-0005", timestamp=ds["timestamps"][5], value=35,
+                    available_at="2026-01-01T04:00:00Z", revision_id="r"), **override}]
+    try:
+        result = analyze(body)
+    except ValueError as exc:
+        assert reason in str(exc)
+        return
+    assert result.dataset_results[0].status == "failed" and reason in result.dataset_results[0].error
+
+
+def test_reset_starts_whole_timestamp_cohort_and_cannot_split_it():
+    body = request()
+    body["config"]["reset_points"] = [body["datasets"][0]["timestamps"][2]]
+    item = analyze(body).dataset_results[0]
+    assert item.status != "failed" and item.data_quality["event_reset_positions"] == [2]
+    body["config"]["reset_points"] = [3]
+    item = analyze(body).dataset_results[0]
+    assert item.status == "failed" and "cannot split" in item.error
+
+
+def test_pointwise_relationships_align_logical_events_and_expose_source_ids():
+    body = request()
+    other = deepcopy(body["datasets"][0])
+    other.update(id="quantity", units="calls", values=[1]*7, rules=[])
+    body["datasets"].append(other)
+    body["relationships"] = [dict(id="unit-cost", kind="ratio", numerator="cost", denominator="quantity", units="dollars_per_call")]
+    result = analyze(body)
+    relationship = result.relationship_results[0]
+    assert relationship.status == "completed"
+    assert relationship.lineage[-1].source_event_ids == {"cost":"call-0006", "quantity":"call-0006"}
+    assert relationship.assessments[-1].event_id == "call-0006"
+    other["event_ids"][-1] = "different-call"
+    failed = analyze(body).relationship_results[0]
+    assert failed.status == "failed" and "matching logical event IDs" in failed.error
+
+
+def test_python_schema_cli_and_mcp_retain_identity_contract():
+    body = request()
+    direct = analyze(body).model_dump(mode="json")
+    assert semantic(analyze_series(body["datasets"][0], body["config"]).model_dump(mode="json")) == semantic(direct)
+    root = Path(__file__).parents[1]
+    jsonschema.validate(body, json.loads((root/"schemas/request-1.1.schema.json").read_text()))
+    jsonschema.validate(direct, json.loads((root/"schemas/result-1.1.schema.json").read_text()))
+    jsonschema.validate({"request":body}, TOOLS[0]["inputSchema"])
+    cli = subprocess.run([sys.executable,"-m","anomalyzer","analyze","-","--input-format","json","--format","json"],
+        input=json.dumps(body), text=True, capture_output=True, timeout=30)
+    assert cli.returncode == 0, cli.stderr
+    assert semantic(json.loads(cli.stdout)) == semantic(direct)
+    def call(name, **arguments):
+        result = handle(dict(jsonrpc="2.0", id=1, method="tools/call", params=dict(name=name, arguments=arguments)))["result"]
+        assert not result["isError"]
+        return result["structuredContent"]
+    compact = call("analyze", request=body)
+    assert compact["data_quality"][0]["event_identity_contract"] == "event-identity-v1"
+    findings = call("list_findings", run_id=compact["run_id"], limit=10)
+    cases = [f for f in findings["findings"] if f["kind"] == "case"]
+    assert len(cases) == 2
+    for case in cases:
+        detail = call("get_evidence", run_id=compact["run_id"], finding_id=case["id"])
+        assert {s["event_id"] for s in detail["source_samples"]} == {case["event_id"]}
+    bad = deepcopy(body)
+    bad["datasets"][0]["event_ids"][1] = bad["datasets"][0]["event_ids"][0]
+    with pytest.raises(jsonschema.ValidationError):
+        jsonschema.validate(bad, TOOLS[0]["inputSchema"])
+
+
+def test_tied_event_results_and_mcp_views_remain_bounded():
+    sizes = []
+    for count in (192,384):
+        values = [10+(i%3-1)*.1 for i in range(count)]
+        values[-1] = 40
+        body = request(values, [i//3 for i in range(count)])
+        body["config"].update(training_size=12,calibration_size=12)
+        result = analyze(body)
+        assert result.status == "completed"
+        sizes.append(len(result.model_dump_json().encode()))
+        compact = summarize(result.model_dump(mode="json"))
+        assert len(json.dumps(compact).encode()) < 15_000
+        assert len(result.dataset_results[0].data_quality["samples"]) == count
+        assert len({r["id"] for r in rows(result)}) == len(rows(result))
+    assert sizes[-1] < 2_000_000 and sizes[-1] < 2.3*sizes[0]
+
+
+def test_revision_order_is_deterministic_and_duplicate_revision_delivery_is_rejected():
+    body = request()
+    ds = body["datasets"][0]
+    ds["revisions"] = [dict(event_id="call-0005", timestamp=ds["timestamps"][5], value=value,
+        available_at=f"2026-01-01T0{hour}:00:00Z", revision_id=f"r{hour}") for hour,value in ((4,35),(5,36))]
+    original = analyze(body)
+    ds["revisions"].reverse()
+    reordered = analyze(body)
+    assert rows(original) == rows(reordered)
+    assert [c.model_dump() for c in original.cases] == [c.model_dump() for c in reordered.cases]
+    ds["revisions"].append(deepcopy(ds["revisions"][0]))
+    failed = analyze(body).dataset_results[0]
+    assert failed.status == "failed" and "duplicate event revision" in failed.error
+
+
+def test_trusted_same_time_prefix_and_shared_point_budget():
+    body = request(trusted=True)
+    original = analyze(body)
+    ds = body["datasets"][0]
+    ds["values"].append(50)
+    ds["timestamps"].append(ds["timestamps"][-1])
+    ds["event_ids"].append("trusted-next-call")
+    ds["event_sequence"].append(ds["event_sequence"][-1]+1)
+    assert rows(analyze(body))[:-1] == rows(original)
+    body["config"]["max_points"] = 7
+    with pytest.raises(ValueError, match="max_points"):
+        analyze(body)
+
+
+def test_identified_lag_does_not_invent_an_order_and_lineage_keeps_original_time():
+    body = request()
+    body["datasets"][0]["timestamps"][0] = "2026-01-01T01:00:00+01:00"
+    other = deepcopy(body["datasets"][0])
+    other.update(id="predictor",rules=[])
+    body["datasets"].append(other)
+    body["relationships"] = [dict(id="lag",kind="lagged_response",response="cost",predictor="predictor",lag=1,coefficient=1,units="dollars_per_call")]
+    result = analyze(body)
+    assert result.dataset_results[0].data_quality["samples"][0]["original_timestamp"] == "2026-01-01T01:00:00+01:00"
+    assert result.relationship_results[0].status == "failed"
+    assert "cannot order tied events" in result.relationship_results[0].error
+
+
+def test_corrected_reference_revises_supported_event_without_changing_its_identity():
+    body = request()
+    original = analyze(body)
+    ds = body["datasets"][0]
+    ds["revisions"] = [dict(event_id="call-0002",timestamp=ds["timestamps"][2],value=8,
+        available_at="2026-01-01T04:00:00Z",revision_id="recalibration")]
+    corrected = analyze(body)
+    a, b = original.cases[0], corrected.cases[0]
+    assert a.id == b.id and a.revision_id != b.revision_id
+    points = [a for a in corrected.dataset_results[0].assessments if a.basis == "statistical_baseline"]
+    assert all(a.baseline["known_input_availability_cutoff"] == "2026-01-01T04:00:00+00:00" for a in points)
+
+
+def test_entity_scope_separates_case_and_evidence_identity():
+    body = request()
+    customer = analyze(body)
+    body["datasets"][0]["entity"] = {"customer":"synthetic"}
+    application = analyze(body)
+    assert {c.id for c in customer.cases}.isdisjoint(c.id for c in application.cases)
+    assert {e["id"] for e in rows(customer)}.isdisjoint(e["id"] for e in rows(application))
diff --git a/tests/test_release.py b/tests/test_release.py
index 1a60e93..cee7cc0 100644
--- a/tests/test_release.py
+++ b/tests/test_release.py
@@ -138,3 +138,23 @@ def fake_gh(*args):
     monkeypatch.setattr(publisher, "gh", fake_gh)
     publisher.publish(dist, "v0.2.1", "a" * 40, finalize=True)
     assert calls[-1] == ("api", f"repos/{publisher.REPOSITORY}/releases/42")
+
+
+def test_draft_visibility_delay_retries_reads_without_creating_twice(monkeypatch, tmp_path):
+    dist = tmp_path / "dist"
+    dist.mkdir()
+    monkeypatch.setattr(publisher, "verify", lambda *args: None)
+    replies = iter([None,None,None,{"id":42,"draft":True,"assets":[]}])
+    monkeypatch.setattr(publisher, "find_release", lambda tag: next(replies))
+    delays, calls = [], []
+    monkeypatch.setattr(publisher.time, "sleep", delays.append)
+    def fake_gh(*args):
+        calls.append(args)
+        if args[0] == "api" and "/commits/" in args[1]:
+            return json.dumps({"sha":"a"*40})
+        assert args[:2] == ("release","create")
+        return ""
+    monkeypatch.setattr(publisher,"gh",fake_gh)
+    publisher.publish(dist,"v0.3.0","a"*40)
+    assert delays == [2,2]
+    assert sum(call[:2] == ("release","create") for call in calls) == 1