Repository navigation
Conversation
…sed port The test closed one server, then started another on port 0. The OS can hand the just-closed port to the new server, so the "nothing listening" address answered. Start the answering server first. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Thread transfer impact✅ Thread transfer remains within every enforced ceiling.
Other scenariosAll 64 scenario measurements match All 64 scenario measurements
Baseline: Scenario and decoded snapshot size10 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.
Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed. |
|
No actionable comments were generated in the recent review. Walkthrough
Merge Risk:
|
Problem
PeerLinks.test.ts> "linking an address with nothing listening says so" starts a server, closes it, then starts a second server on port 0. The OS can give the second server the port that was just closed. Then the "nothing listening" address answers, and the test fails:This happened on 2026-11-21 on
fix/antigravity-null-skill-manifest, which passed on rerun at the same SHA. It happened before on 2026-11-15 (#23531). Neither diff touches this test.Change
Start the answering server first, then open and close the refused one. The closed port can no longer go to the server that answers. This changes only the test.
Scope
Test-only flake fix from the daily CI health pass. The product behavior is correct.
Verification
PeerLinks.test.tspass in 3 of 3 runs.vp run check:changedpasses.Made by Claude Opus 5.5 in Claude Code (T3 Code daily CI health pass).
🤖 Generated with Claude Code