You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
feat(git-shipping): loop Copilot review through receiving-code-review - #169
After opening a PR, the agent now gets it reviewed by Copilot code review and settles every comment on its own until the review stops producing fixes, instead of handing each round back to the user.
Evidence
git-shipping: open PR
→ request @copilot → wait for its review of the head commit
→ receiving-code-review (unattended), per unresolved thread and per finding in the review body:
independent accept → fix, reply, resolve
otherwise → reply with evidence, resolve, list in final report
repeat → keep earlier verdict unless its code or evidence changed
→ commit and push fixes; any fixes and < 3 reviews? repeat : stop and report
This PR ran the loop on itself: review 1 left 2 threads (both accepted and fixed), review 2 left 1 finding only in its body (accepted and fixed), and review 3 reported 0 open findings.
Impact
Copilot never reads replies, so pushed-back items reach a human only through the final report.
Each review costs Copilot AI credits; the 3-review cap bounds it.
intent(review-loop): ship a PR and settle Copilot review without the user in the loop
decision(non-accept): reply with evidence and resolve instead of stopping, listing them in the final report
decision(termination): stop when a review yields no fix commits, or after three reviews
learned(copilot-review): Copilot never reads replies and may re-raise settled comments
rejected(commands): spelling out gh commands; agents already know them
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Unattended fixes may implement held accepted items without clarification
skills/receiving-code-review/SKILL.md:39
The unattended instruction to "fix accepted items" does not preserve the preceding paragraph's hold on items that share an assumption with unclear feedback. A held item can still have an accept verdict, so this branch can lead to implementing a dependent fix without clarification; the final-report rule also omits held accepted items. Explicitly limit unattended fixes to independent accepted items, and report held items with their blocking dependency.
learned(copilot-review): findings on unchanged code appear only in the review body, not as threads
learned(copilot-review): REST comment author is Copilot; GraphQL thread author is copilot-pull-request-reviewer
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
intent(skill-size): keep SKILL.md bodies under 500 words
decision(split): move PR body format, review loop, AI reviewer rules; core flow stays in SKILL.md
Co-Authored-By: Claude Opus 5.5 (1M context) <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
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
After opening a PR, the agent now gets it reviewed by Copilot code review and settles every comment on its own until the review stops producing fixes, instead of handing each round back to the user.
Evidence
This PR ran the loop on itself: review 1 left 2 threads (both accepted and fixed), review 2 left 1 finding only in its body (accepted and fixed), and review 3 reported 0 open findings.
Impact
🤖 Generated with Claude Code