test(server): keep provider fixture directories host-portable

[juliusmarminge]

The Claude capabilities probe can abort before its child has fully exited. On Windows, keeping that child’s cwd inside the scoped fixture makes cleanup fail with EBUSY. The Codex collaboration fixture also hard-coded /tmp, which does not exist on Windows.

Give the Claude child a separate temporary cwd with retrying final cleanup, and use the host temporary directory for Codex collaboration sessions. The provider behavior and assertions are unchanged.

Part of the Windows test-suite stack rooted at #9564; the manual Windows lane comes from #9538.

Originally prepared with Claude Fable 5 in Claude Code; finalized with GPT-5.6 Sol in T3 Code.

---

Note

Low Risk
Test-only fixture and cleanup changes; no production code paths or assertion changes.

Overview
Makes two provider integration tests reliable on Windows by fixing how temporary working directories are created and torn down. Production provider behavior and test assertions are unchanged.

Claude capabilities probe SDK test: The fake child’s workspace cwd is no longer nested inside the scoped fixture directory. It is created as its own temp folder and removed in a finalizer via node:fs/promises rm with retries (instead of Effect scheduling, which would stall under TestClock). That avoids EBUSY when the probe aborts the SDK before the child exits and still holds the directory as cwd.

Codex collab runtime integration tests: Every makeCodexSessionRuntime call that used a hard-coded /tmp cwd now uses os.tmpdir(), so sessions run against the host temp path on Windows and Unix alike.

^{Reviewed by Cursor Bugbot for commit 8170118. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	13.5 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	57.0 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	8	—	21	✅
Claude	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	57.8 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	9	—	21	✅

Baseline: unavailable · PR result: 8170118 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 113.8 KiB
• Claude decoded thread snapshot: 114.5 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — The runtime changes are confined to test cleanup and host-temporary-directory portability, with no product request-path impact. However, the PR adds a file-level nodeBuiltinImport:off static-analysis suppression, so human review is required.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

No code changes detected at 8170118. Prior analysis still applies.

^{You can add or adjust custom eligibility rules. Learn more.}

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{Bugbot Autofix is ON, but a cloud agent failed to start.}

^{Reviewed by Cursor Bugbot for commit bc8fff5. Configure here.}