ShipfoxHQ/shipfox / .agents/skills/fix-ci
fix-ci
VerifiedDiagnose and fix failing CI checks on a pull request, then commit and push the fix. Trigger on /fix-ci, fix CI, CI is red, checks are failing, or fix the failing build on a PR.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 2.3 KB |
| agents/openai.yaml | config | 140 B |
SKILL.md
View on GitHub---
name: fix-ci
description: "Diagnose and fix failing CI checks on a pull request, then commit and push the fix. Trigger on /fix-ci, fix CI, CI is red, checks are failing, or fix the failing build on a PR."
---
# Fix CI
Find why a pull request's checks fail, fix the cause, and push the fix.
## Input
`/fix-ci [pr]`
`[pr]` is a pull request number or URL. Without it, omit `<pr>` below to use
the pull request for the current branch:
```bash
gh pr view <pr> --json number,url,headRefName
```
Use the returned `number` in every later command. If the current branch is not
`headRefName`, run `gh pr checkout <number>`.
## Process
1. If checks are still running, wait for them. CI can outlast a shell command
timeout, so run the command in the background or run it again until it
exits:
```bash
gh pr checks <number> --watch --interval 30
```
2. List the failed checks:
```bash
gh pr checks <number> --json name,bucket,link --jq '.[] | select(.bucket == "fail")'
```
A GitHub Actions `link` has the form
`.../actions/runs/<run-id>/job/<job-id>`. For other checks, such as Vercel,
open the link to read the failure. If nothing failed, report that and stop.
3. Read the failure from the job log. For E2E failures, download the
diagnostics artifact before you form a theory:
```bash
gh run view <run-id> --job <job-id> --log-failed
gh run download <run-id> -p 'e2e-diagnostics*' -D .context/ci/<run-id>
```
4. Reproduce the failure locally with the matching `turbo` task and package
filter. Read the
[local development and release workflow](../../../docs/guides/local-development-and-release-workflow.md)
to pick the task.
5. Fix the cause. Do not skip, weaken, or delete a test to make it pass.
6. If the failure is unrelated to the diff, such as a known flaky test or an
infrastructure error, rerun the failed GitHub Actions jobs instead of
changing code. Rerun only when `gh run view <run-id> --json attempt` reports
attempt 1. For other checks, report the failure instead:
```bash
gh run rerun <run-id> --failed
```
7. Commit the fix through the `commit` skill and push with `git push`.
## Report
Report each failed check with its cause and outcome: fixed in a commit, rerun
as unrelated, or not fixed with the reason. Include the commit hashes pushed.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |