Quick Start
Get Ferret reviewing your pull requests in under 5 minutes — no infrastructure required.
Zero-infrastructure start
The default config uses local file storage for RAG and requires only an Anthropic API key. You can add Qdrant, Pinecone, or other backends later.
Prerequisites
- A GitHub or GitLab repository
- An API key for at least one AI provider (Anthropic, OpenAI, Gemini, etc.)
- Optional for RAG: Ollama installed locally
Step 1 — Add the GitHub Actions workflow
Create the file below in your repository. Ferret will run automatically on every pull request.
on:pull_request:types: [opened, synchronize]jobs:ferret-review:runs-on: ubuntu-latestpermissions:pull-requests: writesteps:- uses: actions/checkout@v4with:fetch-depth: 0- run: |curl -L https://github.com/Arunachalamkalimuthu/merlin-ai-code-review/releases/latest/download/merlin-linux-amd64 -o ferretchmod +x ferret && ./ferret reviewenv:GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
Step 2 — Set repository secrets
Go to your repository → Settings → Secrets and variables → Actions and add:
| Secret | Value | Required |
|---|---|---|
ANTHROPIC_API_KEY | Your Claude API key (sk-ant-...) | Yes (or another provider) |
GITHUB_TOKEN | Provided automatically by Actions | Yes |
Step 3 — (Optional) Add ferret.toml
Drop ferret.toml in your repo root to customise the AI provider, focus areas, and more. Without it, Ferret uses sensible defaults.
[ai]provider = "anthropic"model = "claude-sonnet-4-6"max_tokens = 4096temperature = 0.2[review]focus = ["bugs", "security", "style", "performance"]max_comments = 30chunk_lines = 200reflect = false
Step 4 — Open a pull request
Push a branch and open a PR. Ferret will trigger automatically and post inline comments within a few seconds. You can also trigger commands by commenting on the PR:
@ferret /review@ferret /describe@ferret /security@ferret /ask Is this change thread-safe?
GitLab CI
See the Platforms page for a complete GitLab CI example and instructions for Bitbucket, Azure DevOps, and Gitea.
Next steps
- Configuration reference — all ferret.toml options
- Slash commands — full list of available commands
- RAG pipeline — index your codebase for context-aware reviews
- Autonomous agent — set up Slack/Discord integration