CI/CD Fix Expert Skill

Overview

This Claude Code skill automatically analyzes and fixes CI/CD pipeline failures, including build errors, test failures, linting issues, and formatting problems across different project types.

Features

• Auto-Detection: Automatically detects project type (Foundry, Next.js, Node.js, Rust, etc.)
• Error Analysis: Parses CI logs to identify root causes
• Targeted Fixes: Applies minimal, focused fixes
• Verification: Confirms fixes work before committing
• Safety First: Maintains existing patterns and security

How It Works

Automatic Activation

The skill automatically activates when:

• CI/CD pipeline fails
• User mentions "CI failure", "build error", or "test failure"
• User asks to fix failing workflows
• Analyzing GitHub Actions logs

Fix Process

1. Identify Failure: Get and parse CI logs
2. Detect Project: Determine project type and tools
3. Categorize Error: Build, test, lint, format, or type error
4. Apply Fix: Make minimal targeted changes
5. Verify: Run local checks before committing

Supported Project Types

Project Type	Build	Test	Lint	Format
Foundry/Solidity	forge build	forge test	solhint	forge fmt
Next.js	next build	jest	eslint	prettier
Node.js/TS	npm build	npm test	eslint	prettier
Rust	cargo build	cargo test	clippy	cargo fmt

Error Categories

Build Errors

• Missing imports
• Type mismatches
• Syntax errors
• Missing dependencies

Test Failures

• Assertion failures
• Timeout issues
• Setup problems

Linting Issues

• Code style violations
• Unused variables
• Import ordering

Formatting Issues

• Whitespace
• Indentation
• Line endings

Type Errors

• Missing types
• Wrong types
• Nullable handling

Usage Examples

Example 1: Fix Failed CI

"The CI is failing, can you fix it?"

The skill will:

1. Get the failure logs from the latest run
2. Identify the error type
3. Apply appropriate fixes
4. Verify and commit

Example 2: Specific Run

"Fix the CI failure in run #12345"

Example 3: After Viewing Logs

"I see type errors in the build, please fix them"

Maintenance Rules

DO:

• Fix only what's broken
• Maintain existing patterns
• Keep security best practices
• Verify fixes locally
• Make minimal changes

DON'T:

• Refactor unrelated code
• Add new features
• Change test behavior unnecessarily
• Introduce breaking changes
• Skip verification

Output Format

After fixing, provides a summary:

## CI Fix Summary

### Error Type
[Build/Test/Lint/Format/Type]

### Root Cause
[Explanation]

### Fixes Applied
1. [Fix 1]
2. [Fix 2]

### Files Modified
- file1.ts
- file2.ts

### Verification
- [x] Build passes
- [x] Tests pass

Integration with CI/CD

Use as composite action for automatic fixes:

name: Auto-fix CI

on:
  workflow_run:
    workflows: ["CI"]
    types: [completed]

jobs:
  fix:
    if: ${{ github.event.workflow_run.conclusion == 'failure' }}
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4  # Caller provides setup
      - run: npm ci
      - uses: teliha/dev-workflows/.github/actions/fix-ci@main
        with:
          failed_run_id: ${{ github.event.workflow_run.id }}
          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
          github_token: ${{ secrets.GITHUB_TOKEN }}

Safety Checks

Before pushing:

• All tests pass locally
• Build succeeds
• Linting passes
• Formatting correct
• No new warnings
• Minimal changes
• No security issues

Related Commands

• /fix-ci - Explicitly invoke this skill
• /fix-lint - Focus on linting issues only

Limitations

• Cannot fix fundamental design issues
• May not understand complex business logic
• Requires access to CI logs
• Should be reviewed before merge

Contributing

To improve this skill:

1. Add new error patterns to recognize
2. Add project-specific fixes
3. Improve verification steps
4. Test with real CI failures