Skip review on release PRs; widen the reviewer tool allowlist - #414
Merged
Conversation
The review failed on the v1.19.0 release PR with 20 permission denials and "Reached maximum number of turns (40)". Two causes, both fixed. Release PRs are the wrong thing to review. A dev -> main PR aggregates commits that were each already reviewed on their own PR, so re-reviewing the whole release adds nothing - and the diff is large enough to reliably exhaust the turn budget, which then puts a red X on the release itself. Now skipped. The allowlist was still too narrow. --allowedTools REPLACES the default tool set rather than adding to it, so anything omitted is denied at runtime - and a denied call still consumes a turn. A short list therefore burns the budget and the review dies having posted nothing, which is how both #399 (10 denials) and this one (20) failed. Added the ordinary read-only exploration tools plus TodoWrite, and raised the cap to 60. Everything on the list is read-only except the two comment-posting tools. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The review check failed on the v1.19.0 release PR with 20 permission denials and
Reached maximum number of turns (40). Two separate causes.1. Release PRs are the wrong thing to review
A
dev->mainPR aggregates every commit since the last release - each of which was already reviewed on its own PR. Re-reviewing the whole release adds nothing, and the diff is large enough to reliably exhaust the turn budget. The result is a red X on the release itself, which is the worst possible place for a spurious failure.Now skipped when head is
devand base ismain.2. The allowlist was still too narrow
--allowedToolsreplaces the default tool set rather than adding to it, so anything omitted is denied at runtime - and a denied call still consumes a turn. A short list therefore burns the budget and the review dies having posted nothing.That is the same failure mode twice now: #399 died with 10 denials, this one with 20. Fixing it by adding one or two tools at a time is how I got here, so this adds the ordinary read-only exploration set rather than guessing again:
Read, Grep, Glob, TodoWrite,gh pr comment/diff/view,gh api,git diff/log/show/status,ls, cat, head, tail, wc, find, rgEverything on that list is read-only except the two comment-posting tools. Turn cap raised 40 -> 60.
Note
v1.19.0 was merged with this check red. That was deliberate:
reviewis not a required status check (onlybuild-and-testandcheck-versionare), both required checks passed, and every commit in the release had already been reviewed on its own PR - including the three real bugs the reviewer caught on #411.🤖 Generated with Claude Code