gh-respond-cross-pr
Skill1 starsChanged 4 months agoMITView on GitHub
Respond to review comments across multiple related PRs. Use when the user asks to "handle review comments across PRs", "respond to cross-PR reviews", or "fix feedback on all PRs".
In this repository4 other files it gives its agents.
- explain-codeSkill
- gh-review-cross-prSkill
- gh-review-prSkill
- gh-review-respondSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 45
If a PR is already approved, STOP and ask the user before committing or modifying commits.
- Distinctive: line 47
Human reviewers: ALWAYS create a NEW commit.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Handle review comments across multiple related PRs that belong together.
Arguments: <PR1> <PR2> [PR3...] [--context <issue-url>]
Example: /gh-respond-cross-pr org/frontend#45 org/backend#112 --context org/planning#8
The --context argument is optional and can be a GitHub issue URL or reference. If provided, use it to understand the overall goal and better distinguish real issues from bot noise.
Steps
- Read all unresolved comments across all PRs from all sources:
- Inline review comments:
gh api repos/OWNER/REPO/pulls/NUMBER/comments - Review-level comments:
gh api repos/OWNER/REPO/pulls/NUMBER/reviews(body text submitted with each review) - PR-level comments:
gh api repos/OWNER/REPO/issues/NUMBER/comments
- Inline review comments:
- Identify cross-PR context: Understand how the PRs relate. A comment on one PR may be best addressed by a change in another PR.
- Filter bot noise: Bot reviewers (e.g. Copilot) review each PR in isolation and may flag things that only make sense in the context of the full change set. Identify and flag these:
- Missing imports/definitions that exist in a sibling PR
- "Unused" code that is consumed by a sibling PR
- Naming suggestions that would break consistency with sibling PRs Mark these as dismissible with a short explanation referencing the related PR.
- Present a unified summary to the user: group comments by PR, showing reviewer, file, line, and feedback. Flag bot noise separately.
- Categorize each comment (same as
gh-review-respond):- Fix needed: Real issue, propose a fix. Note which PR the fix belongs in.
- Discussion: Needs user input.
- Disagree/Won't fix: Acknowledge concern, explain reasoning, offer to revisit.
- Bot noise: Cross-PR false positive, dismiss with explanation.
- Ask the user to confirm which comments to fix, discuss, decline, or dismiss.
- Apply fixes in the correct repos. Some fixes from a comment on one PR may need to be applied in another PR's repo.
- Reply to each comment on the appropriate PR:
gh api repos/OWNER/REPO/pulls/NUMBER/comments \ -X POST \ -F in_reply_to=COMMENT_ID \ -f body="Reply text"- For bot noise: reply with a brief explanation (e.g. "This is defined in the sibling PR org/backend#112").
- Resolve all addressed conversations in a single batched GraphQL mutation per PR.
- Stage and commit fixes in each repo:
- If a PR is already approved, STOP and ask the user before committing or modifying commits. New or modified commits invalidate the approval and require re-review.
- Bot reviewers: squash into existing commits with
git commit --fixupandgit rebase --autosquash. - Human reviewers: ALWAYS create a NEW commit. NEVER squash for human reviews. Tell the user to push each repo when ready.
Rules:
- Cross-PR awareness is the primary value. Always consider the full change set before acting on a comment.
- Be constructive and respectful.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

