pr-followup
Skill4 starsChanged 6 days agoMITView on GitHub
Run the review chain for the PR's lane immediately after a PR is opened: /review alone for a pipeline PR, the built-in code-review then /review for an app PR, nothing for docs, release and sync PRs. Invoke right after a PR is created, or manually against an existing PR.
- Commits and pushes
3 other projects use a version of this.
Adopted by
Most recently taken up by
- raditia/CraftKittheir own version, first seen Nov 8, 2026
- LuisUrrutia/skillstheir own version, first seen Oct 18, 2026
- prpande/skillstheir own version, first seen Apr 25, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- LuisUrrutia/skills★ 2same name
- prpande/skills★ 1same name
- raditia/CraftKit★ 1same 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 repository16 other files it gives its agents.
- benchmarkSkill
- bugfixSkill
- designSkill
- deterministic-pr-gatesSkill
- featureSkill
- gatesSkill
- parallel-reviewSkill
- pipeline-reviewSkill
- planSkill
- releaseSkill
- reviewSkill
- specSkill
- statusSkill
- sync-workflowSkill
- testSkill
- trim-contextSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 0 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
0
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 17
Use the built-in code-review skill, not the code-review:code-review plugin.
- Very specific: line 41
Small pipeline PRs: when the lane is pipeline and the PR changes fewer than 30 lines (additions plus deletions, small_pr in …
- Very specific: line 48
Only the app lane (which also covers a PR touching app and pipeline paths) runs both, because on pipeline diffs the second reviewer doubled …
- Very specific: line 59
Run /review <PR>. It posts the verdict and updates the PR body's Review: line.
“Common” is measured: kept by at least 6 in 10 of the 3 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 Followup Agent
Runs the review chain for the PR's lane immediately after a PR is opened, and records each
result where the review-evidence CI check reads it (scripts/check_review_evidence.py). A pipeline PR gets /review alone.
An app PR gets the built-in code-review (at medium effort) and then /review. The
docs, release and sync lanes get neither.
On an app PR, code-review runs first so its fixes land before the /review
rounds; a fix after an APPROVED verdict would otherwise need /review --confirm.
Use the built-in code-review skill, not the code-review:code-review plugin.
The plugin starts 10-20 subagents per run (Haiku triage, 5 Sonnet reviewers, a
Haiku scorer per issue); from 2026-09-28, when it became the default here, it
made each PR's review several times more expensive. /review's isolated
subagent already covers design and rule compliance.
Note: code-review is a Claude Code skill and may not be invocable at all —
either because it is unavailable in this session, or because this coding
agent has no such skill mechanism. Either way, step 3
below catches the invocation error and continues to reporting rather than
halting the whole command; you'll still need to run an equivalent review
yourself before merging.
Trigger
Invoked right after gh pr create succeeds, or manually against an existing
PR: /pr-followup 71 or /pr-followup fix/some-branch.
Process
- Run
python3 scripts/check_pr_lane.py --git origin/<base> --head-branch <head> --base-branch <base>. The script prints the lane:app,pipeline,docs,releaseorsync. For thedocs,releaseandsynclanes, report the lane and stop: they need neither step below. For thepipelinelane, or any PR that also touches pipeline paths (anapp-lane PR still owes the pipeline lane'smotivating_incident, perevidence_for_changeinscripts/check_pr_lane.py), make sure the PR body has a non-emptyMotivating incident: <what went wrong, with a link or date>line (ornone (<reason>)); add it if missing, sincereview-evidencefails without it. Small pipeline PRs: when the lane ispipelineand the PR changes fewer than 30 lines (additions plus deletions,small_prinscripts/pipeline_lanes.json) and touches none of itsexclude_paths(the files that enforce the rules: hooks, gates, review,scripts/check_*,review_verdict.py, the install scripts, lane configs, workflows), skip/reviewtoo: the gates, the lane check andreview-evidencedecide, andreview_verdictis not required. Report "small pipeline PR, no model review" and stop. Count withgit diff --shortstat origin/<base>...HEAD. Lane routing: apipelinePR gets one reviewer, so skip steps 2–3 and go to step 4. Only theapplane (which also covers a PR touching app and pipeline paths) runs both, because on pipeline diffs the second reviewer doubled the token cost and added design suggestions rather than defects (2026-10-05: the same 4-line Gate 0 fix got 7 inline comments across a project PR and pragma #100). applane only: run the built-incode-reviewonce:code-review <PR> medium --comment. Fix any issue it posts, commit and push. Then add or replace one line in the PR body with the head SHA it reviewed:code-review: <comment URL> at <sha>, orcode-review: no issues at <sha>when it posted nothing.applane only: if the step 2 invocation errors (e.g.Unknown skill), don't stall. Print⚠️ code-review couldn't be invoked here (unavailable, or unsupported on this agent) — run an equivalent review yourself before merging.and continue.- Run
/review <PR>. It posts the verdict and updates the PR body'sReview:line. Stop at CHANGES REQUESTED until the blocking issues are fixed./reviewruns one round by default and a second only to confirm blocking fixes. - Report the results (one for a
pipelinePR, two forapp) and the lane.
/test is not in this chain: it now runs between /feature and /gates.
Done when
Each result the lane needs is reported and recorded in the PR body (or the fallback warning printed). Do not merge — per AGENTS.md/CLAUDE.md's Merge rule, the required checks decide mergeability and the user merges.
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.

