address-pr-feedback
Skill0 starsChanged 4 months agoMITView on GitHub
One-shot pass to address PR review comments: fetch threads, apply validated fixes, lint/test affected files, commit, push, reply, and offer to resolve. Use when the user asks to address review comments on a PR or points to a PR link/number. For continuous monitoring until merge-ready, use pr-babysitter instead.
- Deletes or force-pushes
- Commits and pushes
6 other projects use a version of this.
Adopted by
Most recently taken up by
- SDSLeon/lightcodetheir own version, first seen Nov 1, 2026
- Porabuild/Poracodetheir own version, first seen Nov 1, 2026
- ericmasiello/skillstheir own version, first seen Oct 20, 2026
- codemonkey85/PKMDS-Blazortheir own version, first seen Oct 8, 2026
- JeffreyFerreiras/ai-skillstheir own version, first seen Oct 5, 2026
- kyto64/aidd-template-scaletheir own version, first seen Sep 16, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- codemonkey85/PKMDS-Blazor★ 240same name
- Porabuild/Poracode★ 121same name
- SDSLeon/lightcode★ 121same name
- JeffreyFerreiras/ai-skills★ 1same name
- ericmasiello/skills★ 0same name
- kyto64/aidd-template-scale★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository40 other files it gives its agents.
- architectureSkill
- code-reviewSkill
- code-structureSkill
- competitive-briefSkill
- contract-guardSkill
- debugSkill
- deep-workSkill
- dependency-upgradeSkill
- deploy-checklistSkill
- diagnoseSkill
- documentationSkill
- explain-pr-step-by-stepSkill
- feedback-checkpointSkill
- grill-meSkill
- grill-with-docsSkill
- handoffSkill
- improve-codebase-architectureSkill
- incident-responseSkill
- memory-managementSkill
- observabilitySkill
- plan-criticSkill
- plan-from-rfcSkill
- pr-babysitterSkill
- prototypeSkill
- queue-checkSkill
- rebase-onto-mainSkill
- releaseSkill
- security-reviewSkill
- setup-engineering-skillsSkill
- squash-and-pushSkill
- synthesize-researchSkill
- system-designSkill
- tddSkill
- tech-debtSkill
- testing-strategySkill
- to-issuesSkill
- to-prdSkill
- triageSkill
- verification-before-completionSkill
- zoom-outSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 5 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Remove unused components or exports flagged by reviewers.
- Distinctive: line 32
When a reviewer asks "do we have a component for this?", search the codebase for an existing shared component before writing a new one.
- Very specific: line 33
If a comment touches an externally consumed surface, apply the contract-guard skill before changing it.
- Very specific: line 37
Run lint only on the changed files, the way the repo's pre-commit hook would.
- Distinctive: line 60
Ask the user whether addressed comments should be marked resolved.
- Very specific: line 88
Always derive OWNER, REPO, PR_NUMBER at runtime from gh.
- Distinctive: line 89
Don't run the full test suite for a small change — affected files only.
- Distinctive: line 90
Never resolve a thread whose fix wasn't validated.
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Address PR Feedback
A single pass over a PR's review comments: fix what's valid, reply to everything, push.
1. Get PR context
Derive the repo and PR number at runtime — never hard-code them:
gh repo view --json nameWithOwner -q .nameWithOwner
gh pr view --json number -q .number # PR from current branch, if not given explicitly- Fetch review comments:
gh api repos/<OWNER>/<REPO>/pulls/<PR_NUMBER>/comments --paginate - Fetch PR details:
gh pr view <PR_NUMBER> --json title,body,files - Summarise for the user: which comments need code changes, which are discussion-only.
2. Apply code changes
- Address each comment that requires a code change — validate first (see the
pr-babysittervalidation gate: confirmed defect vs preference; behavior impact understood). - Remove unused components or exports flagged by reviewers.
- When a reviewer asks "do we have a component for this?", search the codebase for an existing shared component before writing a new one.
- If a comment touches an externally consumed surface, apply the
contract-guardskill before changing it.
3. Lint
Run lint only on the changed files, the way the repo's pre-commit hook would. Check package.json scripts or the pre-commit config (e.g. lefthook.yml, .husky/) for the right command. Fix errors before proceeding.
4. Run affected tests
Find the test command in the repo's package scripts. Run only the tests affected by the changes — not the full suite — following the repo's test conventions.
5. Commit and push
Stage only the files you changed. Commit with a clear message in the repo's convention. Push. If asked to rebase: git fetch origin main, git rebase origin/main, then git push --force-with-lease.
6. Reply to PR comments
Post a reply to each addressed comment:
gh api repos/<OWNER>/<REPO>/pulls/<PR_NUMBER>/comments/<COMMENT_ID>/replies \
-X POST -f body="<message>"Thank reviewers, confirm what was done ("Done — switched to the shared component. Thanks!"), keep the tone friendly. For declined suggestions, explain the reasoning with evidence.
7. Offer to resolve threads
Ask the user whether addressed comments should be marked resolved. If yes:
- Fetch review thread node IDs (variableized GraphQL):
gh api graphql -f query='
query($owner: String!, $repo: String!, $pr: Int!) {
repository(owner: $owner, name: $repo) {
pullRequest(number: $pr) {
reviewThreads(first: 100) {
nodes { id isResolved comments(first: 1) { nodes { databaseId body } } }
}
}
}
}' -f owner="<OWNER>" -f repo="<REPO>" -F pr=<PR_NUMBER>- Match threads to the comments replied to by
databaseId, then resolve each:
gh api graphql -f query='
mutation($threadId: ID!) {
resolveReviewThread(input: { threadId: $threadId }) { thread { isResolved } }
}' -f threadId="<THREAD_NODE_ID>"Notes
- Always derive
OWNER,REPO,PR_NUMBERat runtime fromgh. - Don't run the full test suite for a small change — affected files only.
- Never resolve a thread whose fix wasn't validated; reply with analysis instead.
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.

