[maschwenk]

test server 1 has been red on every fork main push. storageCleanup.test.ts › "compares the checked-out branch without a full status" points a worktree's .git at /missing-git-directory and expects cleanup to keep it as "Repository or branch changed". git 2.56, which the hosted ubuntu runners have now, rewords that error from not a git repository to gitfile does not point to a valid repository, so cleanup reported the worktree as failed with the raw git line instead. so its real behavior, not a flaky test. anyone on git 2.56 with a worktree whose gitdir is gone gets a failed cleanup entry instead of a kept one

this is a straight port of the upstream fix, which applied cleanly. julius closed the earlier attempt in favor of it, but it isnt merged upstream yet:

• fix(server): git 2.56 broken-gitfile worktrees count as non-repositories again pingdotgg/t3code#33423
• fix(server): Git 2.56 errors for a broken .git file count as not a repository pingdotgg/t3code#26359

isNonRepositoryGitStderr in GitVcsDriverCore.ts now matches both wordings and is exported, and storageCleanup.ts and GitManager.ts call it instead of keeping their own inline copies. tradeoff is we're still string matching git's stderr, so a future reword breaks it again, but both cases exit 128 so there's nothing better to key on. other gitfile errors like invalid gitfile format still count as failures, and a test row covers that

checked against a real git 2.56.0 (homebrew bottle unpacked into a temp dir): the test fails on main with the exact CI error and passes with this change, and it still passes on git 2.53

written by Claude Opus 5.5 in Claude Code

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ The exact PR base did not have a successful artifact. Baseline uses the latest successful main measurement shown below.
⚠️ This PR changes transfer ceilings:

• Codex Live turn WebSocket decoded: 29.3 KiB → 24.4 KiB
• Claude Live turn WebSocket decoded: 29.3 KiB → 24.4 KiB

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	+33 B (+0.6%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	+33 B (+2.7%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	21.5 KiB	+674 B (+3.2%)	24.4 KiB	✅
Codex	Live turn messages	2	1	−1 (−50.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	+32 B (+0.6%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	+32 B (+2.6%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.9 KiB	+686 B (+3.2%)	24.4 KiB	✅
Claude	Live turn messages	2	1	−1 (−50.0%)	8	✅

Baseline: cdc3e41 · PR result: db88846 · 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: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

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