CI/CD Integration

Integrating Claude Code with CI/CD pipelines.

Claude Code can run in GitHub Actions for automated quality checks. Here's how we integrate company-standard skills into CI.

1.1 Basic Pattern

yaml
# .github/workflows/claude-review.yml
name: AI Code Review

on:
  pull_request:
    branches: [main, develop]

jobs:
  ai-review:
    runs-on: ubuntu-latest
    permissions:
      pull-requests: write
      contents: read
    steps:
      - uses: actions/checkout@v4
      
      - name: Run Security Review
        uses: anthropics/claude-code-action@beta
        with:
          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
          prompt: |
            Run the security review skill on the changes in this PR.
            Report any critical issues as PR comments.
            Use the skill at .claude/skills/security-review/SKILL.md

1.2 Zone-Based CI Enforcement

Zone A — Advisory only:

yaml
# Runs code-review skill, posts findings as comments
# Does not block merge on AI findings alone

Zone B — Blocking:

yaml
# Runs security-review skill
# Fails the CI check if CRITICAL severity issues are found
# Requires human resolution before merge

Zone C — Don't automate Claude Code in CI

1.3 Using Skills in CI

Reference shared skill files directly:

yaml
- name: AI Quality Gate
  uses: anthropics/claude-code-action@beta
  with:
    anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
    prompt_file: .claude/skills/security-review/SKILL.md
    additional_prompt: "Focus on the files changed in this PR only."