pr-comments
Skill1 starsChanged 4 months agoMITView on GitHub
Triage the review feedback on a pull request — human reviewers and AI/bot reviewers alike — into a judged task list. Use when asked to "pull the comments from my PR", "look at the review feedback", "work through the PR comments", or after opening a PR that receives an automatic AI review. Collects every unresolved comment, judges each against repo facts, and verdicts it ACT / DISMISS / HUMAN with cited evidence. It does not reply on GitHub and does not apply fixes.
19 other projects use a version of this.
Adopted by
Most recently taken up by
- skyfox675/agents-skillstheir own version, first seen Nov 9, 2026
- domengabrovsek/agent-configtheir own version, first seen Nov 8, 2026
- domengabrovsek/claudetheir own version, first seen Nov 8, 2026
- ieuanign/skillstheir own version, first seen Nov 3, 2026
- flavio-portela/skillstheir own version, first seen Nov 1, 2026
- K-log/skillstheir own version, first seen Nov 1, 2026
Used with
Projects with it also use worktree, jira and 5 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
MicrosoftSkillworktree
Create, reuse or clean up a dedicated git worktree for working on a pull request or a private branch. Use from any PR-scoped workflow that needs an isolated checkout to keep current code and build output untouched.
Also in 3 projects that use this
MicrosoftSkilljira
Jira issue workflows for search, issue updates, transitions, comments, field discovery, and interactive credential setup via the Jira REST API. Use when you need to configure Jira access, search with JQL, inspect an issue, create or update work items, move an issue between statuses, post comments, or discover required fields for issue creation.
Also in 3 projects that use this
simo97Skillwrite-a-skill
Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.
Also in 3 projects that use this
AWSSkillship
Finalize dev work on an issue branch — update docs, push, and open a PR.
Also in 4 projects that use this
DockerSkillreview-pr
Review one or more incoming Docker documentation pull requests as a maintainer. Independently validate technical claims, assess editorial fit and information architecture, choose a verdict, and draft exact inline or PR-wide feedback behind a confirmation gate. Use for requests such as "review PR 123", "is this PR correct?", "does this information belong here?", "validate this PR", or "help review backlog PRs". Do not use to maintain or fix a PR you own; use maintain-pr for that.
Also in 3 projects that use this
GoogleSkillprototype
Build a throwaway prototype to answer a design question. Use when the user wants to sanity-check whether a state model or logic feels right, or explore what a UI should look like.
Also in 3 projects that use this
Other versions19 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 19
- ccch1mneyyy/dsh-TUI★ 7.8ksame name
- nudgebee/nudgebee★ 525same name
- vstorm-co/agenticos★ 315same name
- BrOrlandi/my-claude-skills★ 22same name
- domengabrovsek/agent-config★ 22same name
- domengabrovsek/claude★ 22same name
- skyfox675/agents-skills★ 11same name
- hdcola/skills★ 4same name
- tomazb/agent-skills★ 4same name
- ieuanign/skills★ 2same name
9 more not listed
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 repository21 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ai-commentsSkill
- ai-draftSkill
- ai-nextSkill
- ai-postSkill
- call-traceSkill
- draft-prSkill
- dto-mappingSkill
- handoffSkill
- hum-commentsSkill
- jira-dcSkill
- meeting-summarySkill
- plan-from-issueSkill
- plan-nextSkill
- plan-reviewSkill
- plan-saveSkill
- pr-statsSkill
- review-sessionSkill
- skill-builderSkill
- validator-generatorSkill
- templateSkill
The file
We split this skill into the rules it gives an agent
1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.
0
Common
Most of the 18 edited versions we read keep these.
1
Distinctive
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 29
After opening a PR, once its reviews (human or automatic AI) have posted
“Common” is measured: kept by at least 6 in 10 of the 18 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
PR Comments: turn review feedback into judged, actionable tasks
You are working through the review feedback on a pull request. The failure mode this skill prevents is treating feedback — especially AI-reviewer feedback — as either gospel or noise: each comment gets a grounded verdict — valid and worth acting on, or incorrect and safe to dismiss — backed by evidence from the repo, never by vibes. You triage; you do not fix, and you do not post.
When to use this skill
- "Pull the comments from the PR" / "look at the comments on my PR" / "go through the review feedback"
- After opening a PR, once its reviews (human or automatic AI) have posted
When NOT to use this skill
- No PR exists yet — review the branch locally instead; this skill triages posted feedback.
- The user wants the fixes applied — triage first with this skill; applying accepted items is a follow-up outside it.
- Summarizing a PR's discussion for status purposes — just read it; this skill is for triage.
Steps
- Resolve the PR. From a given number/URL, or the current branch (
gh pr viewwith no args). If no PR can be resolved, ask and stop until answered. Collect the feedback. Four sources, all of them:
- Review bodies and states:
gh pr view <n> --json reviews - Inline review comments:
gh api repos/{owner}/{repo}/pulls/<n>/comments— GET only; this skill reads, it never writes (never-X POST/PATCH/DELETE). - Conversation comments:
gh pr view <n> --json comments - Thread resolution (so "skip resolved" is real, not assumed — the REST endpoints above don't expose it):
gh api graphql -f query='{repository(owner:"<owner>",name:"<repo>"){pullRequest(number:<n>){reviewThreads(first:100){pageInfo{hasNextPage endCursor} nodes{isResolved comments(first:50){nodes{databaseId path line}}}}}}}'— map every comment'sdatabaseIdin each thread to that thread'sisResolved(a reply inside a resolved thread must be dropped too — don't stop at the first comment); ifpageInfo.hasNextPage(a PR with >100 review threads), re-run the query withafter:"<endCursor>"to page through the rest.
Default scope is all unresolved feedback — human and AI/bot alike; narrow to bot-authored only when the user asks for just the AI review. Drop threads whose
isResolvedis true (and any already replied to with a decision), using the resolution map above — never guess resolution. If this leaves zero items, distinguish "reviewed, nothing flagged" from "the review hasn't posted yet": when the repo runs an automatic AI review on PR open and no bot review is present inreviewsyet, tell the user it likely hasn't posted and to re-run shortly — report a clean/empty triage only when at least one review actually exists.- Review bodies and states:
- Build the task list. One numbered item per distinct piece of feedback: quote (or tightly paraphrase) the comment, its file/line anchor, and its thread id. Merge duplicates that restate the same point.
- Judge each item. Check the claim against reality: read the flagged file at the flagged line, grep the repo for the conventions or symbols the comment invokes, and read the surrounding code rather than the hunk alone. When a judgment needs intent context beyond the code — what the change was supposed to do — read the PR description and any linked issue/ticket via
ghbefore verdicting; if intent is still genuinely ambiguous, that item is HUMAN, not a guess. Verdicts:- ACT — the comment is right; state what the fix is (don't apply it).
- DISMISS — the comment is wrong or moot; state the evidence (
file:line) that refutes it. - HUMAN — judgment call above this skill's pay grade (product intent, team preference); say what the human needs to decide.
- Present the triage in the output format, then stop. Do not reply to threads, resolve threads, or edit files. If the user wants accepted items applied or dismissals posted as replies, that happens after this skill, at their direction.
Output format
PR #<n> — <title>
Feedback items: <total> ACT: <n> DISMISS: <n> HUMAN: <n>
1. [ACT] <file:line> — "<comment gist>"
Fix: <what to change, one line>
2. [DISMISS] <file:line> — "<comment gist>"
Evidence: <file:line / fact that refutes it>
3. [HUMAN] <file:line> — "<comment gist>"
Decide: <the question for a human>
Suggested order: <ACT items grouped by file, cheapest verification first>Rules
What to do
- Every verdict carries evidence. ACT names the concrete fix; DISMISS cites the file/fact that refutes the comment; HUMAN states the open question. "Seems right" is not a verdict.
- Be willing to dismiss. Review feedback — AI feedback especially — is advisory; a grounded DISMISS with evidence is as valuable as an ACT.
- Collect from all three comment surfaces — review bodies, inline comments, conversation — inline-only misses half the feedback.
What NOT to do
- Never apply fixes or push commits from this skill — the deliverable is the judged task list.
- Never post to GitHub (no replies, no thread resolution, no reviews) — what goes back to the PR is the user's voice.
- Don't re-litigate resolved threads or items the user already decided — triage only open feedback.
Format discipline
- The triage table exactly as above, verdict tags first on each line, then stop. No per-comment essays — one line of reasoning each.
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.

