fix(antigravity): avoid spawning runtimes for health checks

[rynfar]

Antigravity health checks launch a PyInstaller runtime repeatedly, which can leave large extraction folders after forced shutdown. Resolve installation metadata without starting the runtime, and isolate each actual ACP runtime in its own scoped temp directory.

Adopts upstream 8c18b5bb21a5349fbadba8c77456ea34c41b7339 (#12008), preserving original authorship and credit to VitaCodez and ariszz described in the upstream PR. Pylon intentionally excludes the blanket startup sweep: another driver/server may still own files beneath the same profile. Cleanup removes only the current runtime's acquired directory after child teardown. Abrupt termination or failed Windows cleanup can leave directories; system temp is untouched. The probe returns only the installation metadata consumed by provider health, not synthetic protocol capabilities.

Validation: 61 focused driver, auth-environment and provider tests pass; server typecheck, scoped lint/formatting and diff checks pass. Tests cover repeated non-spawning probes, resolution failure, separate runtime directories, teardown cleanup and preservation of unowned files. Platform environment mapping is tested; no real Windows Antigravity binary was exercised. No wire/client behavior changes; session authentication and model refresh still use actual runtimes.

Part of #585. Frozen upstream head 6d1d549441be84f19696ab59ed7e2fbf305280d4; global cursor unchanged. No release or installation.

Implemented with GPT-6 in Codex.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
pylon-marketing	Ready	Preview	Sep 17, 2026 5:37am UTC

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.9 KiB	13.9 KiB	+14 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.2 KiB	7.2 KiB	−6 B (−0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.7 KiB	6.7 KiB	+20 B (+0.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	58.0 KiB	58.0 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	14.1 KiB	14.1 KiB	+23 B (+0.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.2 KiB	7.2 KiB	+4 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.8 KiB	6.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	59.7 KiB	59.7 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	9	9	0 (0.0%)	21	✅

Baseline: 00b2ffc · PR result: 7f57442 · 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: 115.6 KiB
• Claude decoded thread snapshot: 116.3 KiB

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