[linkerlin]

Summary

未完成 run 上，readonly_smoke 点名的 swarm_doctor 在软缺口（soft_continue=true）时现在也给出 next_action=swarm_solidify 和 preserve 三元组。只读冒烟之后按 next_action 走的宿主会去固化，而不会因为缺这个键、又看到「循环没被挡住」再去 tick。gate_soft_continue 与 session_not_failed 仍然缺席。本调用不结算这条 run，不写 Parent baseline，不执行 --fix，也不放宽固化门。

On an unfinished run, a soft-continue swarm_doctor report (the other tool named by readonly_smoke) now sets next_action=swarm_solidify and the preserve triad. A host that smokes and then follows next_action solidifies instead of ticking because the action was missing and the loop looked unblocked. gate_soft_continue and session_not_failed stay absent. The call does not settle the run, write a Parent baseline, run --fix, or relax the solidify gate.

What changed

• 仅当存在未完成 run、soft_continue=true、且没有硬 blockers 时，swarm_doctor 增加 pending_solidify、next_action=swarm_solidify、readonly_smoke（settles_run=false）和 preserve_pending=true / abandon_run=false / action_required=swarm_solidify。hint 先写现在调用 swarm_solidify，不要 tick，固化前不要执行 fixes。
• 不写 gate_soft_continue。缺席不是 false，避免把严格 false 读成整段会话失败（DEBUG #67）。不写 session_not_failed：那个旗标只属于 tick 重定向，不是 doctor 的健康位。
• 没有未完成 run 时 doctor 形状不变。硬阻断（例如不是 git 仓库，ok=false）不另给 next_action。
• 工具描述与返回体同一口径。DEBUG #61 的宿主目录缓存没有做假修复。
• 不改第一次 tick 重定向（仍是 ok=true、retryable=false、没有 error、没有 loop_fuse、软缺口上 gate_soft_continue 缺席）。不改保险丝（ok=false、loop_fuse=pending_redirect_loop）。

How to test

1. uv run pytest tests/test_mcp_host_usability_pending_smoke.py tests/test_mcp_server.py tests/test_mcp_protocol.py tests/test_mcp_host_usability_p0.py tests/test_mcp_host_usability_p1.py tests/test_mcp_host_usability_p2.py tests/test_mcp_host_usability_polish.py tests/test_swarm.py -q --tb=line
   • Expected: 153 passed.
   • test_pending_doctor_on_a_soft_gap_names_solidify fails on unmodified master (KeyError: pending_solidify). The hard-blocker case stays ok=false with no next_action.
2. uv run ruff check src tests && uv run ruff format --check src tests && uv run mypy src
   • Expected: ruff clean, mypy Success: no issues found in 353 source files.

Risk

Low. The soft-continue doctor envelope gains a solidify steer only while a run is unfinished. ok, soft_continue, and loop_ready stay as the open-box report computed them. Hard failures do not gain next_action. Solidify gate, skip_validation, Parent baseline, and val seal are unchanged.

Harness/evaluator governance

Upstream governance surface: MCP swarm_doctor return while a run is unfinished (soft-continue only). Not solidify, selector, mutation, or the frozen gate.
Downstream EvoX impact: N/A -- existing read-only tool return, no bridge or contract change.
Rollout-local scope: host-visible doctor smoke only; boot, status, and the tick redirect envelope are unchanged.
Promotion boundary: N/A -- no proposal/rollout/default promotion.
Evaluator mismatch sets: N/A
Non-regression evidence: tests/test_mcp_host_usability_pending_smoke.py pins the soft-continue steer (next_action=swarm_solidify, preserve triad, gate_soft_continue and session_not_failed absent, no baseline file) and the hard-blocker path (no new action).
Fix-severity review: low
Owner approval: N/A
Security boundary: no new data, tool, host, network, or secret surface. Doctor still calls doctor(fix=False).
Rollback: revert this commit.
Live promotion: no
Autonomous evaluator self-editing: no

Self-check

• No new runtime dependencies without a clear justification in "What changed".
• Tests added or updated to cover the new behavior; gates pass locally: uv run pytest (modules listed above), uv run ruff check src tests, uv run ruff format --check src tests, uv run mypy src.

  

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.