Setup Goose Action

A GitHub Action that installs and caches the Goose AI agent for CI/CD workflows. Goose is an open-source AI coding agent by Block. This composite action downloads the binary, caches it across runs, and adds it to PATH.

Available on the GitHub Marketplace

[!IMPORTANT]
Prompt Injection Risk: When AI analyzes user-controlled input (git diffs, code comments, commit messages), malicious actors can embed instructions to manipulate output. This applies to ANY AI tool, not just Goose or this action.

For production use, see Security Patterns below for three defensive tiers (tool output analysis, manual approval, trusted-only execution).

Features

• Caching: Automatically caches the Goose binary for faster subsequent runs
• Version Pinning: Install specific Goose versions for reproducible builds
• Lightweight: Composite action with no external dependencies

Usage

# Recommended: Get latest v1.x updates automatically
- uses: clouatre-labs/setup-goose-action@v1

# Pin to exact version
- uses: clouatre-labs/[email protected]

# Custom Goose version
- uses: clouatre-labs/setup-goose-action@v1
  with:
    version: '1.13.0'

Current default Goose version: See action.yml

Prerequisites

1. Get an API key from your chosen provider: Supported Providers

2. Add it as a repository secret:

   • Go to Settings > Secrets and variables > Actions
   • Click New repository secret
   • Name it (e.g., GEMINI_API_KEY, OPENAI_API_KEY, ANTHROPIC_API_KEY)

3. Configure in your workflow by mapping your secret to Goose's expected environment variable (see Security Patterns below)

Quick Start: Tier 1 (Maximum Security)

name: Secure AI Analysis
on: [pull_request]

permissions:
  contents: read

jobs:
  analyze:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@v6

      - name: Lint Code
        run: pipx run ruff check --output-format=json . > lint-results.json || true

      - name: Setup Goose CLI
        uses: clouatre-labs/setup-goose-action@v1

      - name: AI Analysis
        env:
          GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
        run: |
          mkdir -p ~/.config/goose
          cat > ~/.config/goose/config.yaml << 'EOF'
          GOOSE_PROVIDER: google
          GOOSE_MODEL: gemini-2.5-flash
          keyring: false
          EOF
          
          echo "Summarize these linting issues:" > prompt.txt
          cat lint-results.json >> prompt.txt
          goose run --instructions prompt.txt --no-session --quiet > analysis.md

      - name: Upload Analysis Artifact
        uses: actions/upload-artifact@v5
        with:
          name: ai-analysis
          path: analysis.md

Inputs

Outputs

Security Patterns

This action supports three security tiers for AI-augmented CI/CD:

• Tier 1 (Maximum Security): AI analyzes only tool output (JSON), never raw code. See workflow
• Tier 2: AI sees file stats, requires manual approval. See workflow
• Tier 3: Full diff analysis, trusted teams only. See workflow

Safe Pattern: AI analyzes tool output (ruff, trivy, semgrep), not raw code.

Unsafe Pattern: AI analyzes git diffs directly, which is vulnerable to prompt injection.

Read the full explanation: AI-Augmented CI/CD blog post

See SECURITY.md for reporting vulnerabilities.

Supported Platforms

Note: This action only supports Linux runners. macOS runners have a 10x billing multiplier on GitHub Actions, and Goose executes prompts and tool calls with nothing platform-specific.

How It Works

1. Checks cache for Goose binary matching the specified version and platform
2. If cache miss, downloads Goose binary from official GitHub releases
3. Extracts binary to ~/.local/bin/goose
4. Adds binary location to $GITHUB_PATH
5. Verifies installation with goose --version

Troubleshooting

Binary not found after installation

Ensure you're using the action before attempting to run goose:

- uses: clouatre-labs/setup-goose-action@v1
- run: goose --version  # This will work

Unsupported version

Check available versions at Goose Releases. Ensure the version exists and has pre-built binaries.

Contributing

See CONTRIBUTING.md for the full contribution guide, including commit signing requirements, coding standards, and the PR process.

License

MIT. See LICENSE.

Related

• AI-Augmented CI/CD: 3-tier security model for AI code review in CI/CD pipelines
• Goose: Official Goose repository
• Goose Documentation
• GitHub Actions Documentation
• Setup Kiro Action: Similar action for Kiro CLI (AWS-native, SIGV4 auth)
• Setup Q CLI Action: Similar action for Amazon Q Developer CLI

Acknowledgments

Built by clouatre-labs for the Goose community. Not officially affiliated with Block or the Goose project.