Claude Code GitHub Actions

What is Claude Code GitHub Actions?

It is a GitHub Action (anthropics/claude-code-action@v1) that runs Claude Code inside your existing CI/CD pipelines. Once installed, anyone on your team can type @claude in a PR comment or issue and Claude will respond with analysis, code changes, or even a new pull request.

Unlike the standalone Code Review feature (which automatically scans every PR), GitHub Actions gives you full control — Claude only acts when you ask it to, or when a workflow you write triggers it.

How it works

The action supports two modes — interactive mode (responds to @claude mentions in comments) and automation mode (runs immediately using a prompt you define in the workflow file). The mode is detected automatically based on your configuration.

Example — @claude in a PR comment

A developer asks Claude to review JWT security. Claude reads the code, finds an issue, and posts a detailed comment in seconds.

Setup

Common use cases

Type these in any PR or issue comment once the action is installed.

Automation mode — run without @claude

You can also trigger Claude automatically using any GitHub event — a scheduled cron job, every PR opened, or a push to main. Use the prompt parameter to give Claude its instructions directly in the workflow file.

name: Daily Summary
on:
  schedule:
    - cron: "0 9 * * 1-5"   # 9 AM every weekday

jobs:
  summarise:
    runs-on: ubuntu-latest
    steps:
      - uses: anthropics/claude-code-action@v1
        with:
          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
          prompt: "Summarise yesterday's merged PRs and open issues. Post the summary as a new issue."
          claude_args: "--max-turns 5"

name: Auto PR Review
on:
  pull_request:
    types: [opened, synchronize]

jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: anthropics/claude-code-action@v1
        with:
          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
          prompt: "Review this pull request for correctness, security, and code style. Post findings as review comments."
          claude_args: "--max-turns 5"

Key configuration options

* Not required when using Amazon Bedrock or Google Vertex AI — use their respective credential parameters instead.

Customise behaviour with CLAUDE.md

Create a CLAUDE.md file at your repository root to define coding standards, review criteria, and project-specific rules. Claude reads this file before acting and follows the guidelines in every response.

# Project Guidelines

## Code style
- Use TypeScript strict mode
- Prefer named exports over default exports
- All async functions must handle errors explicitly

## Review focus
- Flag any use of `eval()` or `new Function()`
- Require input validation at all API boundaries
- All database queries must use parameterised statements

## PR conventions
- One logical change per PR
- Update CHANGELOG.md for user-facing changes

Best practices

Troubleshooting

Quick Reference