test(deploy): 補齊預設匯入探測成功與失敗動態覆蓋

[monkey1sai]

變更摘要

補齊 Start-HostNativeKitManager 預設 ImportProbeFn 的真實 Python subprocess 成功與失敗動態覆蓋，並保留既有 timeout／descendant cleanup 案例。

本 PR 是使用者指定的 WP-20260811-001-import-probe-dynamic follow-on test slice。Issue #493 已由 PR #502 的 timeout coverage 關閉；本 PR 不重新關閉 #493，也不宣稱 #489、#513 或完整 work package 已完成。

修改原因

目前 origin/main 已動態驗證預設 probe timeout，但 success 與 non-zero failure 仍缺少真實 child-process 證據。這個缺口可能讓 import fastapi, uvicorn 的執行順序、失敗阻擋或 service-start wiring 回歸而未被 focused test 發現。

主要變更

• 修改 scripts/tests/test-host-native-launcher.ps1，production code 零變更。
• Success fixture 以暫存 fastapi.py／uvicorn.py shim 與 ordered marker 證明兩個 import 都在真實 child interpreter 執行，且 service 僅啟動一次。
• Failure fixture 讓 fastapi.py 寫 marker 後 raise，證明 documented blocker 被傳遞、service 不啟動、uvicorn 不會被 import。
• 既有 timeout fixture 保留，繼續驗證 bounded failure 與 parent/child PID cleanup。
• 所有 test-only environment variables 均在 finally 還原。

Change Classification

Item	Result
Change lane	G
Behavior contract changed	no
Requirement source	issue #493（已關閉）＋使用者指定 follow-on work package WP-20260811-001-import-probe-dynamic

AI Coding Governance

Item	Result
Linked issue	Refs #493；已由 #502 關閉，本 PR 不自動關閉任何 issue
Requirement source	issue #493（已關閉）＋使用者指定 follow-on work package WP-20260811-001-import-probe-dynamic
CODEOWNERS / owner review	monkey1sai-blip exact-head approval required；尚未取得
GitNexus evidence	risk_level=UNKNOWN — GitNexus 1.6.9 未索引 PowerShell symbol，且 current linked worktree 無 exact-path index；以單檔 test-only diff、production blob unchanged、focused runtime、root contracts 與兩份獨立 read-only review 補強
Browser E2E evidence	not applicable — 非 user-facing，frontend/product path 零變更
Agent workflow changed?	no
Required checks expected	CI、Agent Governance、PR Metadata Contract、Governance Base Audit（diagnostic）

驗證方式

Exact base/head：

• Base: c152c79d57fbbcf05276e3c1a9bbbaa3ce0d07c3
• Head: fd886afe4947f2305e49fed149fed0d210881d44

已執行：

• pwsh -NoProfile -NonInteractive -File scripts/tests/test-host-native-launcher.ps1 → ALL PASSED；commit 後 exact head 重新執行通過。
• Mutation self-check：暫時將 production argv 改成只 import fastapi，新 marker 斷言如預期失敗；還原後 focused test 重新通過，production blob 與 HEAD 相同。
• pwsh -NoProfile -NonInteractive -File scripts/verify-all.ps1 -PlanOnly -ChangedPath scripts/tests/test-host-native-launcher.ps1 → affected plan 產生成功。
• repo-local TEMP 下 python -m pytest tests -q -p no:cacheprovider → 483 passed, 9 skipped。
• pwsh -NoProfile -NonInteractive -File scripts/tests/invoke-powershell-static.ps1 → passed。
• pwsh -NoProfile -NonInteractive -File scripts/tests/test-agent-governance-check.ps1 → all assertions passed。
• pwsh -NoProfile -NonInteractive -File scripts/tests/scan-secret-patterns.ps1 → passed。
• pwsh -NoProfile -NonInteractive -File scripts/tests/verify-security-exceptions.ps1 → valid，count=0。
• merge 最新 main 後，node --test scripts/tests/test-openspec-machine-truth.mjs → 24 passed, 0 failed。
• merge 最新 main 後，node --test scripts/tests/test-ai-coding-metrics.mjs → 13 passed, 0 failed。
• git diff --check → clean。

Hosted CI：

• Exact head fd886afe4947f2305e49fed149fed0d210881d44 的 CI workflow 已通過：changed-path classifier、Linux platform adapter、root contracts/fakes、agent governance contracts、PowerShell static 與 secret scan 均為 success；不適用的 service jobs 依 classifier skip。
• Exact head 的 Agent Governance / agent-governance 已通過，包含 OpenSpec machine-truth、PR body evidence、self-referential bootstrap、Windows platform adapter、host-native child-launch regression、remote deploy transport、host-native preflight 與 Windows verification scope。
• Exact head 的 PR Metadata Contract / pr-metadata-contract-diagnostic 已通過。
• Hosted manifest 不會直接執行 focused test-host-native-launcher.ps1；本 PR 不把本機 focused pass 誤報成 hosted required gate。
• 既有 ledger provenance defect 已由 chore(openspec): rebind hifi row to the landed #507 squash #524 合併至 main（c152c79）；本 branch 以 merge commit fd886af 吸收，fresh local machine-truth 24/24 與 hosted Agent Governance 均通過。重複且已成零 diff 的 chore(openspec): rebind migrate-console-to-hifi-design row to the landed #507 squash #523 已關閉為 superseded。
• Governance Base Audit 依 repo 契約是 diagnostic；目前 reason 為缺少 monkey1sai-blip 對 fd886af 的 exact-head approval，會在 review submitted 時重跑。本 PR 不把它繞過或誤報為已通過。

風險與影響

• 僅測試程式變更，不修改 runtime、API、資料結構、環境變數契約、部署流程、migration 或 Webhook。
• 新 fixture 以 PYTHONPATH 與 environment variable 傳遞 sandbox path，避免 PowerShell/Python path escaping；Windows 已實跑，Linux runtime 尚未觀察。
• PR fix(deploy,verify,launcher): mechanism-hardening-2 bundle (#490 #491 #489-B) #513 仍在處理獨立 containment findings 且修改同一測試檔；本 patch 與其目前已觀察內容沒有 textual conflict，但合併排序後 fix(deploy,verify,launcher): mechanism-hardening-2 bundle (#490 #491 #489-B) #513 仍須 merge 最新 main 並重跑 exact-head checks/review。
• chore(openspec): rebind hifi row to the landed #507 squash #524 的 ledger rebind 已透過 merge 最新 main 納入 branch history；PR 對 current main 的實際 diff 仍只有 scripts/tests/test-host-native-launcher.ps1。
• scripts/** 依 repo merge governance 屬 elevated scope；本 PR 不繞過 CODEOWNER 或外部 merge authority。

回滾方式

若需撤回，revert 本 PR 的單一 test commit 即可；沒有 production state 或資料需要復原。

後續建議

• Exact-head hosted checks 已完成；目前只等待固定 CODEOWNER review。任何 head 漂移後重新驗證。
• test(deploy): 補齊預設匯入探測成功與失敗動態覆蓋 #525 仍需 monkey1sai-blip 對最終 exact head 提交人工 approval；agent／GitHub App review 只能作證據，不能代替該票。
• 若本 PR 先 merge，將最新 main merge 進 PR fix(deploy,verify,launcher): mechanism-hardening-2 bundle (#490 #491 #489-B) #513，不 rebase／force-push，並重新取得 fix(deploy,verify,launcher): mechanism-hardening-2 bundle (#490 #491 #489-B) #513 的 exact-head gate 與 approval。
• 完整 WP-20260811-001 containment 工作仍由 fix(streaming): contain the Kit process tree on converter timeout (#489 PR-A) #509／fix(deploy,verify,launcher): mechanism-hardening-2 bundle (#490 #491 #489-B) #513 與其 fixpoint 流程處理，本 PR 不宣稱完成。

[chatgpt-codex-connector]

You have reached your Codex usage limits for security reviews. Please try again later.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 3819974f-6a5e-43f3-8492-163838856785

📥 Commits

Reviewing files that changed from the base of the PR and between c152c79 and fd886af.

📒 Files selected for processing (1)

• scripts/tests/test-host-native-launcher.ps1

---

📝 Walkthrough

Walkthrough

The host-native launcher tests now cover successful, failed, and hung Python import probes. They verify marker output, bounded failures, service-start suppression, descendant-process cleanup, and restoration of the marker-file environment variable.

Changes

Host-native import-probe regression coverage

Layer / File(s)	Summary
Import-probe outcomes and cleanup scripts/tests/test-host-native-launcher.ps1	The tests run real Python subprocesses for successful and failed fastapi/uvicorn imports. They verify service-start counts, documented errors, bounded completion, timeout handling, descendant cleanup, and marker-file environment restoration.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Possibly related issues

• launch-time OS containment boundary for host-native child processes (Job Object / cgroup) #522 — Covers related host-native launcher child-process containment and cleanup behavior.

Possibly related PRs

• monkey1sai/AI-BIM-governance#124 — Added the host-native launcher test file extended by this PR.
• monkey1sai/AI-BIM-governance#467 — Expands related host-native launcher and process cleanup regression coverage.
• monkey1sai/AI-BIM-governance#502 — Introduced related import-probe timeout and process-cleanup coverage.

Suggested reviewers: monkey1sai-blip

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly identifies the test scope and the added success and failure coverage for the default import probe.
Description check	✅ Passed	The description directly explains the test-only changes, objectives, validation results, and remaining approval requirement.

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch test/default-import-probe-dynamic-success-failure

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[monkey1sai-blip]

{"kind":"ai-bim-single-owner-approval","version":1,"repo":"monkey1sai/AI-BIM-governance","prNumber":525,"headOid":"fd886afe4947f2305e49fed149fed0d210881d44","baseOid":"c152c79d57fbbcf05276e3c1a9bbbaa3ce0d07c3","action":"merge-elevated"}