What does this skill do, and when should you use it?
This skill is designed to debug or fix failed GitHub Actions checks on GitHub pull requests. It uses the GitHub CLI to inspect check status, run details, and logs, then extracts an actionable failure snippet and summarizes the context. It proposes a fix plan before making changes and implements only after explicit approval. External providers such as Buildkite are out of scope.
Verifies gh authentication; resolves the current branch PR or a supplied PR number or URL; reads GitHub Actions checks, run metadata, and logs; fetches job logs through the GitHub API when a run is still in progress; extracts and summarizes failure snippets; labels non-GitHub Actions checks as external and reports only their details URL; after approval, applies the fix and suggests rerunning tests and gh pr checks.
- A developer needs to investigate failed GitHub Actions checks on the PR for the current branch.
- A maintainer provides a PR number or URL and wants its failing checks inspected.
- A team needs concise, machine-friendly failure context extracted from lengthy CI logs.
- A check is still running and job-level logs must be fetched directly for actionable details.
How do you install this skill?
- Confirm the GitHub CLI authentication scopes before execution, and avoid writing or displaying logs containing tokens, secrets, or personal data.
- GitHub API job logs may be returned as ZIP archives, which this script treats as unparseable; manually verify that logs are genuinely unavailable.
- The repository is marked deprecated; confirm an active maintenance source and compatibility before adoption.
- GitHub service and CLI reachability may be limited from mainland-China networks.
- Shell / CLI
- Network access
- Local filesystem
GitHub CLI (gh)Python
Install the curated skill in Codex with: $skill-installer gh-fix-ci. Restart Codex after installation so it can load the skill. The README does not state an independent license for this skill; check LICENSE.txt in the skill directory.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/openai/skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/.curated/gh-fix-ci" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Inspect this PR’s GitHub Actions failures, summarize the logs, and propose a fix plan; do not modify anything without my approval.
First run gh auth login and ensure the usual repo and workflow scopes are available. Example: python "<path-to-skill>/scripts/inspect_pr_checks.py" --repo "." --pr "123". Add --json for machine-friendly output. A suitable trigger is: “Inspect this PR’s GitHub Actions failures, summarize the logs, and propose a fix plan; do not modify anything without my approval.”
What are this skill's strengths and limitations?
- Covers PR resolution, check inspection, log retrieval, failure summarization, and approval-gated implementation.
- Includes inspect_pr_checks.py with optional JSON output.
- Keeps scope focused on GitHub Actions and avoids attempting work with external CI providers.
- Requires an authenticated GitHub CLI and typically repo and workflow scopes.
- Handles GitHub Actions only; external checks such as Buildkite are limited to reporting their details URL.
- The source provides no test suite, cross-platform validation, or concrete repair strategy.
- The repository is deprecated; its README recommends the OpenAI Plugins repository for current Codex skill and plugin examples.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| GitHub PR Check Fixer this page ✓ OpenAI · Official | 47 · Use with care | ★ 28k | 4mo ago | — |
| Playwright DevOps ✓ Microsoft · Official | 44 · Not recommended | ★ 98k | 11d ago | Apache-2.0 |
| React Native GitHub Actions Artifacts | 49 · Use with care | ★ 1.7k | 17d ago | MIT |
| Megatron-Core CI Failure to Issue ✓ NVIDIA · Official | 46 · Use with care | ★ 3.6k | 11d ago | Apache-2.0 |
| CI Failure Fixer ✓ Microsoft · Official | 48 · Use with care | ★ 194k | 11d ago | MIT |
The repository is deprecated; its README recommends the OpenAI Plugins repository for current Codex skill and plugin examples.
How did FollowSkills review this skill?
The skill is evidenced as primarily performing read-only gh queries and log retrieval, asks the user to authenticate when needed, and requires explicit approval before implementing changes. However, it requests repo and workflow scopes without documenting sensitive data in logs, data flows, least-privilege boundaries, or rollback, so points are deducted.
The script includes authentication checks, PR resolution, field-compatibility fallback, run-log/job-log fallback, and explicit error states. It has no committed tests, marks ZIP-formatted API logs as unparseable, and uses coarse failure-marker extraction. Static review cannot exceed 10 points, so points are deducted.
The audience, inputs, GitHub Actions scope, and non-GitHub- Actions boundary are reasonably clear. Applicability is limited by assumptions about the current branch PR, dependence on GitHub CLI and GitHub network access, and no stated Chinese-language or mainland-China reachability support.
The skill provides overview, prerequisites, inputs, quick start, workflow, and bundled-resource documentation, with mostly stable naming and entry points. It lacks a skill version, changelog, named maintenance responsibility, and update path; license status is not confirmed in the supplied material, and the repository README marks the repository deprecated, so points are deducted.
The script can identify failing checks, retrieve run metadata and logs, extract snippets, and emit text or JSON, covering the core diagnostic task. Fix implementation still depends on a subsequent agent workflow, log parsing may be incomplete, and no statically verifiable representative outputs or comparative benefit evidence are supplied, so points are deducted.
The source code, parameters, and failure branches are auditable, but there is no committed test suite, CI coverage, or third-party execution evidence. The conclusion therefore rests on limited static evidence and receives only a low score.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →