Repository navigation
fix: bound Stop hook stdin read so disabled gate cannot hang on Windows (#530) - #544
stantheman0128 wants to merge 1 commit into
Conversation
fs.readFileSync(0) waits for EOF. On Windows Claude Code may leave the pipe open, so the Stop hook hit the 900s timeout even when stopReviewGate was false. Check config from cwd/env first and use a timed stdin reader (openai#530). Co-authored-by: Cursor <[email protected]>
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5bb448044d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if (!preliminaryConfig.stopReviewGate) { | ||
| const input = await readHookInput(DISABLED_GATE_STDIN_TIMEOUT_MS); |
There was a problem hiding this comment.
When the hook is launched without CLAUDE_PROJECT_DIR or from a different working directory than the project in the hook payload, this early return trusts the config for cwdGuess. The subsequent short read may discover input.cwd, but the branch only logs running tasks and returns, so a project whose payload cwd has stopReviewGate: true can stop without running the review gate. This regresses the previous behavior, which loaded config after reading input.cwd; re-check the actual workspace's config before deciding the gate is disabled.
Useful? React with 👍 / 👎.
…eview prompt via stdin New lib/hook-input.mjs readHookInput(): streams stdin with a deadline (2 s default, CODEX_HOOK_STDIN_TIMEOUT_MS for tests), a 1 MiB byte limit and a StringDecoder across chunk boundaries. EOF parses the whole input; at the deadline a buffer that is already a complete JSON object is accepted (only the EOF was late), anything else is `timeout`; over the limit is `overflow` without parsing; bad JSON is `invalid-json`. Listeners and the timer are removed and stdin destroyed, so the hook exits right after deciding. Stop hook: any read error blocks as before (fail-closed), except a timeout with nothing received while the workspace gate (CLAUDE_PROJECT_DIR / cwd) is off, which allows (openai#530/openai#544). With the gate on, a timeout blocks naming "hook input did not arrive". The review prompt now reaches the companion via `task --prompt-stdin`, not argv, lifting the Windows/Linux argv limits. SessionStart/SessionEnd hook: reads with a 1 s deadline before the 12 s SessionEnd budget starts; on any read error it prints one stderr line and returns without cleanup. The hooks.json timeout test now asserts 15 s > budget + read. readStdinIfPiped (companion): readSync loop that keeps accumulated bytes across EAGAIN on a non-blocking stdin (up to 50 x 20 ms consecutive) and throws on exhaustion instead of returning a partial prompt (openai#120/openai#247/openai#123/openai#150/openai#165). Co-authored-by: stantheman0128 <[email protected]> Co-authored-by: tmchow <[email protected]> Co-Authored-By: Claude Fable 5.1 <[email protected]>
Summary
Fixes #530.
stop-review-gate-hook.mjsusedfs.readFileSync(0), which blocks until stdin EOF. On Windows, Claude Code can leave the write end open after sending the Stop payload, so the hook sat until the 900s hook timeout even whenstopReviewGatewas already false.Changes:
CLAUDE_PROJECT_DIR/ cwd first and take a short timed stdin read when the gate is offreadHookInput(timeoutMs)helper for boundeddata/endcollectionRelated open EAGAIN PRs (#123/#132/#165) target #120 (non-blocking EAGAIN throws). This is a different failure mode: missing EOF with a disabled gate.
Evidence
AI assistance
Assisted by Cursor/Grok. Human author: Stan Shih (@stantheman0128). I reviewed the diff and ran the tests above.