babysit-prs
Skill4 starsChanged 57 days agoMITView on GitHub
Check all open PRs, surface CI failures, and spawn fix agents for red ones
11 other projects use a version of this.
Adopted by
Most recently taken up by
- flopperj/dotfiles-coretheir own version, first seen Nov 5, 2026
- mqmalagris/agent-skillstheir own version, first seen Nov 5, 2026
- ptrandev/claude-skillstheir own version, first seen Oct 29, 2026
- Taxuspt/garmin_mcptheir own version, first seen Oct 17, 2026
- pedrosatin/skillstheir own version, first seen Oct 17, 2026
- kaiitunnz/claude-skillstheir own version, first seen Sep 8, 2026
Other versions11 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 11
- Taxuspt/garmin_mcp★ 1.4ksame name
- brittanyellich/loop-board★ 93same name
- BenjaminG/ai-skills★ 4same name
- flopperj/dotfiles-core★ 3same name
- sabirmgd/terminal-first-goal-loop★ 2same name
- kaiitunnz/claude-skills★ 1same name
- ptrandev/claude-skills★ 1same name
- meirlabs/drain-studio★ 0same name
- mqmalagris/agent-skills★ 0same name
- pedrosatin/skills★ 0same name
1 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 repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-and-stateSkill
- ci-and-releaseSkill
- development-automationSkill
- loggingSkill
- mode-blueprintsSkill
- pr-feedbackSkill
- project-mapSkill
- tui-standardsSkill
- web-frontendSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.
0
Common
Most of the 11 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use the selected assistant's subagent facility and pass the referenced reviewer procedure as its instructions.
- Very specific: line 32
Read advisory Codex findings as well as the enforced verdict.
“Common” is measured: kept by at least 6 in 10 of the 11 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
Babysit the open pull requests so finished work doesn't pile up. Arguments (optional): the user’s arguments — fix to auto-spawn fix agents, otherwise report-only.
Use the selected assistant's subagent facility and pass the referenced reviewer procedure as its instructions. If that facility is unavailable, run the tasks sequentially in their isolated worktrees. Codex does not need to start Claude to follow these procedures.
Survey —
gh pr list --state openthengh pr checks <number>for each. Build a table: PR, branch, CI status (green/red/pending), feedback (frommake pr-feedback PR=<n>), review status, mergeable.The feedback column is the one that used to be missing.
gh pr checkssays whether the machines are happy; it says nothing about the review comments sitting on the PR, which is how work merged past them for months.- Green PRs — a PR is ready only when CI and
pr-feedbackare green. A PR with red feedback is not "ready to merge, pending review" — it is in bucket 4 below. Do NOT merge anything yourself unless the PR carries theauto-mergelabel (thengh pr merge --auto --squashis allowed; the required checks still hold it until the feedback is answered). - Red PRs — for each failing PR, fetch the failing run's log (
gh run view <run-id> --log-failed) and summarize the root cause in one line.- If
fixwas passed: for each red PR, create a headless worktree (make wt-headless NAME=fix-pr-<number>) and spawn thepr-fixersubagent (defined in.agents/reviewers/pr-fixer.md) in the background, passing the branch name, the failure summary, the failing log excerpt, and the worktree path. Its procedure (reproduce → minimal fix → verify → push to same branch) lives in the agent definition. Track the agents, report when they finish, and remove each worktree (make wt-one-rm) once its PR is green. - Otherwise: just report the failures with their causes.
- If
PRs with unanswered feedback — for each PR whose
pr-feedbackstatus is red, runuv run python scripts/pr_feedback.py --pr <n> --jsonand summarize what is open in one line per item.- If
fixwas passed: create a headless worktree (make wt-headless NAME=feedback-pr-<number>) and spawn thepr-respondersubagent (defined in.agents/reviewers/pr-responder.md) in the background, passing the PR number, the branch name, the JSON verdict, and the worktree path. Its procedure — read what was said, decide fix-or-answer per item, reply then resolve — lives in the agent definition, and/pr-feedbackis the same procedure done by hand. Track the agents, report when they finish, and remove each worktree (make wt-one-rm) once its gate is clear. - Otherwise: report the open items and leave them.
Two states here are not work:
pendingmeans the review has not run yet, and "Claude Review never posted a verdict" means the workflow itself did not fire — checkgh run list --workflow "Claude Review"rather than spawning an agent at a diff that is fine.- If
- Stale PRs — flag PRs behind
mainby many commits or inactive for days; suggest/sync-mainon their worktree. - Report — end with a compact status table and the list of actions taken or recommended.
Read advisory Codex findings as well as the enforced verdict. They are returned in advisory_items; a green gate does not mean those findings were addressed.
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.

