pr-fix-ci

Diagnose and fix failing CI checks on pull requests by reading logs and pushing code fixes.

Updated Jul 11, 2026
One-click install
npx skills add https://github.com/danielsuguimoto/skills --skill pr-fix-ci-danielsuguimoto
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: pr-fix-ci
Source: https://github.com/danielsuguimoto/skills/tree/main/skills/pr-fix-ci
Command: npx skills add https://github.com/danielsuguimoto/skills --skill pr-fix-ci-danielsuguimoto

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Failing CI checks block pull requests from merging, and manually digging through check logs to find root causes is slow and error-prone. This Skill automates the loop of reading failure logs, classifying failures, fixing the underlying code, and pushing the fix. ## Core Features & Use Cases - Failure Log Analysis: Reads failing check logs via gh run view --log-failed or fetches rendered output for non-GitHub-Actions checks, extracting the exact error lines, failing files, and failing commands. - Failure Triage: Classifies each failure as code, flaky, infra, or required-status so only real code problems get fixed and transient issues are re-run instead. - Approval-Gated Fixes: In review mode, presents a summary of fixes and local validation results and requires explicit approval before committing and pushing; auto mode pushes without approval. - Use Case: A PR shows three red checks. Invoke this Skill with the PR number to check out the branch, identify that two failures share one root cause in a test assertion, fix it, validate locally, and push after your approval. ## Quick Start Fix the failing CI checks on pull request 142 by reading the logs, fixing the code, and pushing after my approval.

Frequently Asked Questions about pr-fix-ci

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix failing CI checks on a GitHub pull request?▼

Use gh pr checks or gh pr view --json statusCheckRollup to list failing checks, then gh run view <run-id> --log-failed to read the error logs. Fix the root cause in the code, validate locally by re-running the failing command, then commit and push to the PR branch.

How to read GitHub Actions failure logs from the command line?▼

Run gh run view <run-id> --log-failed to see only the failed steps of a workflow run. If that returns nothing, fall back to gh run view <run-id> --log for the full log output.

Should I fix code for flaky or infrastructure CI failures?▼

No. Flaky failures like network timeouts or known-flaky tests should be re-run, not fixed in code. Infrastructure failures such as missing secrets or runner issues should be reported to the user rather than patched in the PR.

Can this Skill push fixes without my approval?▼

Only in auto mode, which commits and pushes without asking. The default review mode requires explicit approval at the approval gate after local validation, and never infers approval from context.

What happens when a failing check cannot be reproduced locally?▼

The Skill states that the check cannot run locally, for example platform-specific builds, and relies on the pushed fix instead. It does not claim CI is green since remote verification has not occurred.