[i-am-sijia]

This pull request updates the model execution environment, improves extension handling, and updates test and baseline data to reflect changes in the model structure and outputs. The most important changes are grouped below.

Model execution and extension handling

• The model is now always executed from the model directory, and the --project .. flag is used to ensure correct path resolution for configs and data. This change is reflected in both the documentation (docs/execution-backends.md) and the model CI script (scripts/model_ci.py).
• Extension handling is improved: the code now explicitly requires the model/extensions directory to exist and consistently uses this path for extension discovery and tracking.

Baseline and test updates

• Baseline outputs in tests/baselines/10000.json and tests/baselines/2000.json are updated to reflect changes in model outputs, including new counts and distributions for persons, tours, trips, and mode shares. Notably, the mode RIDEHAIL is replaced by RIDESOURCE in these outputs.
• The test import path for extensions is updated to use the model subdirectory, ensuring tests import from the correct location.
• A test assertion is updated to check for the new mode name RIDESOURCE instead of RIDEHAIL, matching the model output changes.

[Copilot]

Copilot review overview

🟡 Changes recommended

Extension provenance hashing and documented backend commands still reference the obsolete root-level extension location.

Review effort: Balanced
Findings: 1 · 1

Open (2)

• Hash the loaded extension package in run provenance · New
• Update backend command docs to run from model · New

What changed in this PR

Fixes extension loading and modernizes ActivitySim progress monitoring.

Changes:

• Corrects walk-ability test imports and mode naming.
• Derives monitoring state from resolved run plans and incremental logs.
• Runs CI from model so extensions load correctly.

File	Description
tests/​test_walk_ability.py	Fixes extension path and RIDESOURCE assertion.
src/​lighthouse/​sys_monitor.py	Adds dynamic run-plan and log-based monitoring.
scripts/​model_ci.py	Corrects CI extension loading and working directory.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[i-am-sijia]

@mmilkovits, I updated the baseline distributions in the test folder: tests/baselines/10000.json and tests/baselines/2000.json, corresponding to the "broader" (10,000-household) and "model" (2,000-household) CI tests.

When CI runs these tests, it compares the generated output distributions against the baseline distributions. Differences are reported for review but do not result in test failures.

I ran the following command locally to regenerate and replace the two baseline distribution files.

uv run --locked python scripts/model_ci.py --households 2000 --sharrow off --output model/output_ci_off
uv run --locked python scripts/model_ci.py --households 10000 --sharrow off --output model/output_ci_off