Bump both reviewer pins to ai-review-prompts@533f841#91
Merged
Conversation
Adopts ai-review-prompts#42 — two production-PR-driven fixes from
oauth#89's first real run:
1. log-review-to-ai-review-log.sh now counts findings from the
standardized "N blockers found." line both reviewers' prompts
emit, with header-counting as the fallback. The prior
`^### [0-9]\.` regex missed Claude's evolved inline-comment +
terse-recap pattern; result on oauth#89 was the Claude log
issue titled "no blockers" while 2 inline blockers were posted.
The new regex tolerates optional markdown wrapping
(`**2 blockers found.**` / `_1 blocker found._`).
2. Gemini's continuity preamble is now mandatory when the
prior-findings reference is non-empty. Round 2 on oauth#89
showed Gemini correctly BEHAVING with continuity (dropping the
resolved findings, adding a new one) but not NARRATING it —
the updated comment read like an unrelated fresh review.
Tightened prompt requires the top-level summary line to append
a brief continuity note ("Reviewed; no blockers found. Prior K
resolved by <sha>." or "N blockers found. Prior K resolved by
<sha>; new findings below.") so updates are legible at a
glance on the PR.
_claude-review.yml is byte-identical between ea19009 and 533f841;
bumping in lockstep stays a no-op for Claude behavior but keeps
both callers on the same shared upgrade motion.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Contributor
|
Reviewed; no blockers found. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Adopts HarperFast/ai-review-prompts#42 — two production-PR-driven fixes from oauth#89's first real run.
1. Log script counts from the standardized count line
ai-review-prompts#42 changes
log-review-to-ai-review-log.shto count findings from theN blockers found./Reviewed; no blockers found.line both reviewers' prompts emit, with header-counting (^### [0-9]\.) as the fallback for unknown formats. The new regex also tolerates optional markdown wrapping (**2 blockers found.**/_1 blocker found._).The prior
^### [0-9]\.regex undercounted on Claude's evolved inline-comment + terse-recap pattern. On oauth#89, Claude posted 2 inline blockers + a top-level recap formatted as1. **path:line** — short. See inline., but the ai-review-log title saidno blockers. Now it'll title as2 finding(s) — triage pendingmatching the actual count.2. Gemini continuity preamble now mandatory
Round 2 on oauth#89 showed Gemini correctly behaving with continuity (dropping resolved findings, adding new ones) but not narrating it. Claude's parallel comment said
Reviewed; no blockers found. All three findings from the prior run are addressed in acc5147.Gemini just posted the new finding with no preamble — the updated comment read like an unrelated fresh review.ai-review-prompts#42 tightens the Gemini prompt: when the continuity-reference section is non-empty, the top-level summary line must append a brief continuity note on the same line as the count:
Same-line so the
^[0-9]+ blockers? foundcount-line pattern still matches for fix #1.Pin bump scope
Both reviewer callers move in lockstep.
_claude-review.ymlis byte-identical betweenea19009and533f841; Claude behavior is unchanged._gemini-review.ymlgets the tightened continuity prompt.Test plan
claude-reviewandgemini-reviewworkflows fire on this PR🤖 Generated with Claude Code