ci-watch
Skill0 starsChanged 4 months agoMITView on GitHub
After opening or pushing a PR, watch its CI run to completion and fix any failure — lint, types, tests, build — then re-push until green, instead of leaving a red PR
- Commits and pushes
4 other projects use a version of this.
Adopted by
Most recently taken up by
- jisuzzzz/haechitheir own version, first seen Nov 8, 2026
- cill-i-am/skillstheir own version, first seen Oct 20, 2026
- nubjs/nubtheir own version, first seen Oct 8, 2026
- wilsto/claude-code-starter-kittheir own version, first seen Mar 7, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- nubjs/nub★ 4.4ksame name
- jisuzzzz/haechi★ 17same name
- cill-i-am/skills★ 1same name
- wilsto/claude-code-starter-kit★ 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 repository29 other files it gives its agents.
- adopt-existing-projectSkill
- adrSkill
- branch-conflict-checkSkill
- compound-learningsSkill
- config-auditSkill
- cost-checkSkill
- coverage-ratchetSkill
- doc-syncSkill
- eval-harnessSkill
- flaky-triageSkill
- followup-trackingSkill
- incident-captureSkill
- kit-doctorSkill
- kit-updateSkill
- logging-initSkill
- migration-checkSkill
- new-project-bootstrapSkill
- nightly-auditSkill
- observability-checkSkill
- parallel-workSkill
- perf-budgetSkill
- pr-babysitterSkill
- pre-prSkill
- prompt-regressionSkill
- releaseSkill
- session-handoffSkill
- sync-healthSkill
- using-the-kitSkill
- writing-kit-skillsSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
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 47
Use the systematic-debugging skill.
- Distinctive: line 67
Commit with a message that says what CI caught:
- Very specific: line 83
A flaky test is a finding, not a nuisance: If a test passes on re-run with no code change, log it in the follow-ups file (docs/followups.md …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CI Watch-and-Fix
A PR isn't done when it's opened — it's done when CI is green. Use this skill
immediately after gh pr create or any push to a PR branch. It watches the run,
and if anything fails, diagnoses and fixes it in the same worktree, then
re-pushes and watches again. The pre-pr skill should catch most of this
locally; this is the backstop for the environment-specific failures that only
surface in CI (missing lockfile entry, OS-specific test, cache mismatch).
Profile-driven. Toolchain commands referenced below (lint, types, tests, install) come
from .claude/kit.yaml; the literals shown are the defaults for a Python/uv project.
Work in the PR's worktree the whole time. Never fix CI by editing the trunk_branch
(read from .claude/kit.yaml, default: main).
Workflow
1. Find the run and watch it
branch=$(git branch --show-current)
gh run watch "$(gh run list --branch "$branch" --limit 1 --json databaseId -q '.[0].databaseId')" --exit-status--exit-status makes the command exit non-zero if the run fails, so you can
branch on it. If gh run watch returns success, you're green — report and stop.
gh run watchpolls until completion, so it can block a while. If you'd rather not hold the turn, use the/loopskill orScheduleWakeupto re-check on an interval matched to your CI's typical duration.
2. On failure, get the logs for the failed jobs only
run_id=$(gh run list --branch "$branch" --limit 1 --json databaseId -q '.[0].databaseId')
gh run view "$run_id" --log-failed3. Diagnose before you touch anything
Use the systematic-debugging skill. Read the actual error — don't pattern-match to a fix. Map the failure to its cause:
Reproduce-locally commands below are the defaults; substitute the value from the matching key in
.claude/kit.yamlif your project overrides it:toolchain.lint/toolchain.typecheck/toolchain.test/toolchain.install.
| CI job failed | Reproduce locally (kit.yaml key) | Typical fix |
| lint (ruff) | ruff check src/ tests/ (toolchain.lint) | ruff check --fix, then hand-fix the rest |
| types (mypy) | mypy src/ (toolchain.typecheck) | add/repair annotations; don't # type: ignore to silence |
| tests | pytest <the failing node id> (toolchain.test) | fix the code or the test — establish which is wrong first |
| build | run the build command (toolchain.build) | missing dep in lockfile, bad import path |
| lockfile drift | uv lock / npm install (toolchain.install) | commit the regenerated lockfile |
Reproduce the failure locally first. A fix you can't reproduce failing is a fix you can't verify.
4. Fix, re-run the local gate, push
Apply the fix, then re-run the relevant pre-pr steps locally to confirm.
Commit with a message that says what CI caught:
git commit -am "fix(ci): <what failed> — <what you changed>"
git push5. Watch again
Return to step 1. Repeat until green. If the same job fails a third time on the same cause, stop and report — you're likely misreading the failure, and thrashing wastes CI minutes. Escalate with the logs rather than guessing again.
Guardrails
- Never make CI pass by weakening the check (deleting the assertion, adding a blanket ignore, marking the test skip) unless the check itself is genuinely wrong — and if it is, say so explicitly in the commit and PR body.
- Don't disable a CI job to get green. Fix the cause.
- A flaky test is a finding, not a nuisance. If a test passes on re-run with
no code change, log it in the follow-ups file (
docs/followups.mdby default) rather than silently moving on.
Output
## CI Watch
- PR: #<n>
- Runs observed: <k>
- Failures fixed: <lint / types / tests / build / none>
- Final status: GREEN / STILL RED — <blocking failure + logs>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.

