copilot-review-loop
Skill2 starsChanged 29 days agoMITView on GitHub
Use when the user wants a pull request taken through GitHub Copilot review: requesting a Copilot review, fixing or replying to Copilot review comments, resolving its threads, or looping until Copilot approves. Do not use for general code review or for reviews by people.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- sourcehawk/feature-dev-workflowtheir own version, first seen Nov 15, 2026
- xpepper/pr-review-agent-skilltheir own version, first seen Aug 25, 2026
- moreh-dev/runestheir own version, first seen May 27, 2026
- molon/hunkwisetheir own version, first seen Apr 1, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- molon/hunkwise★ 87same name
- xpepper/pr-review-agent-skill★ 6same name
- sourcehawk/feature-dev-workflow★ 2same name
- moreh-dev/runes★ 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 repositoryOne other file it gives its agents.
- CLAUDE.mdCLAUDE.md
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 4 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use the Artifacts repository's Make targets throughout the review loop.
- Very specific: line 19
Do not load all comments or resolved threads by default.
- Distinctive: line 23
Open thread: Verify the claim, then fix it or reply with evidence that no change is needed.
- Distinctive: line 25
Review summary: Use it as review state, not as a task list.
- Distinctive: line 36
Before editing, search the changed scope for the same cause.
- Very specific: line 61
Run make pr-watch request=1 in the background.
- Very specific: line 65
When the watcher finishes, run make pr-summary and make pr-review-comments.
- Distinctive: line 71
If another round produces actionable comments, repeat the loop.
“Common” is measured: kept by at least 6 in 10 of the 4 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
Copilot review loop
Use the Artifacts repository's Make targets throughout the review loop.
Inspect the current round
Inspect the summary and open threads:
make pr-summary
make pr-review-commentsDo not load all comments or resolved threads by default. Use make pr-review-comments show=all only when the summary and open-thread list disagree or when you need to confirm the state of a thread addressed during this loop.
Classify each result before editing:
- Open thread: Verify the claim, then fix it or reply with evidence that no change is needed.
- Suppressed note: Ignore it unless independent verification finds a material problem in the changed scope.
- Review summary: Use it as review state, not as a task list.
Verify and group
Trace the relevant callers, state changes, contracts, and tests. Do not accept Copilot's explanation after reading only the commented line or local function.
Answer two questions separately:
- Is the claim correct?
- Is the change material and in scope?
Before editing, search the changed scope for the same cause. Group related findings into one patch so the next review does not rediscover the same problem elsewhere.
Address the threads
Make the smallest grouped fix and run the relevant Make checks. Commit and push through Make.
Reply to and resolve every addressed open thread:
make pr-address thread=PRRT_... < notes.mdPass the id exactly as make pr-review-comments prints it. If the environment blocks GraphQL, the listing prints a numeric id instead of PRRT_..., and make pr-address accepts that number. Add pr_num=N when the PR is not on the current branch.
State what you verified, what changed or why no change was needed, and the validation result. A pushed fix does not replace the reply and resolution.
Confirm that no addressed thread remains open:
make pr-review-comments
make pr-summaryRequest the next round
Run make pr-watch request=1 in the background. The request=1 argument ensures that the watcher waits for a new Copilot review instead of accepting the previous round.
Report progress while the watcher is silent. If the wait is interrupted, inspect the current review state before starting another watcher.
When the watcher finishes, run make pr-summary and make pr-review-comments. If historical suppressed notes prevent classification, inspect only the latest review or the specific new comment identified by the watcher.
Stop
Finish when the latest Copilot review recommends approval, no new actionable comments exist, no threads remain open, and the required checks pass. Historical resolved threads and suppressed notes do not block completion.
If another round produces actionable comments, repeat the loop. Stop and report the conflict if the same finding returns after two verified fixes or if resolving it would expand the change's scope.
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.

