# CLAUDE.md

## Common Commands

### Development Commands

- Build/Type check: `bun run typecheck`
- Format code: `bun run format`
- Check formatting: `bun run format:check`
- Run tests: `bun test`
- Install dependencies: `bun install`

### Action Testing

- Test action locally: `./test-local.sh`
- Test specific file: `bun test test/prepare-prompt.test.ts`

## Architecture Overview

This is a GitHub Action that allows running Claude Code within GitHub workflows. The action consists of:

### Core Components

1. **Action Definition** (`action.yml`): Defines inputs, outputs, and the composite action steps
2. **Prompt Preparation** (`src/index.ts`): Runs Claude Code with specified arguments

### Key Design Patterns

- Uses Bun runtime for development and execution
- Named pipes for IPC between prompt input and Claude process
- JSON streaming output format for execution logs
- Composite action pattern to orchestrate multiple steps
- Provider-agnostic design supporting Anthropic API, AWS Bedrock, and Google Vertex AI

## Provider Authentication

1. **Anthropic API** (default): Requires API key via `anthropic_api_key` input
2. **AWS Bedrock**: Uses OIDC authentication when `use_bedrock: true`
3. **Google Vertex AI**: Uses OIDC authentication when `use_vertex: true`

## Testing Strategy

### Local Testing

- Use `act` tool to run GitHub Actions workflows locally
- `test-local.sh` script automates local testing setup
- Requires `ANTHROPIC_API_KEY` environment variable

### Test Structure

- Unit tests for configuration logic
- Integration tests for prompt preparation
- Full workflow tests in `.github/workflows/test-base-action.yml`

## Important Technical Details

- Uses `mkfifo` to create named pipes for prompt input
- Outputs execution logs as JSON to `/tmp/claude-execution-output.json`
- Timeout enforcement via `timeout` command wrapper
- Strict TypeScript configuration with Bun-specific settings