nitpik Documentation
AI-powered code reviews for your team. Bring your own model, bring your own API key.
Getting Started
New to nitpik? Start here:
- Installation — download the binary, install from source, or pull the Docker image.
- Quick Start — run your first review in under two minutes.
- LLM Providers — connect Anthropic, OpenAI, Gemini, or any compatible API.
- GitHub Models — run nitpik on free GitHub-hosted models with
GITHUB_TOKEN, ideal for CI.
Using nitpik
- Diff Inputs — git diffs, file scans, patches, and stdin.
- Reviewer Profiles — built-in specialist reviewers and how to combine them.
- Custom Profiles — write your own reviewer with Markdown and YAML.
- Agentic Mode — let the LLM explore your codebase with tools.
- Output Formats — terminal, JSON, GitHub, GitLab, Bitbucket, and Forgejo.
How It Works
- How Reviews Work — context assembly, multi-agent coordination, and quality post-processing.
- Caching & Prior Findings — content-hash caching and iterative review continuity.
- Project Documentation Context — teach the reviewer your team's conventions.
Security & Privacy
- Security Model — what nitpik sends to LLM providers, subprocess sandboxing, telemetry, and on-disk data.
- Secret Scanning — detect and redact secrets before code reaches the LLM.
- Threat Scanning — detect obfuscated payloads, dangerous APIs, supply chain attacks, and homoglyph tricks.
Deployment
- Configuration Reference —
.nitpik.toml, environment variables, and CLI flags. - CI/CD Integration — GitHub Actions, GitLab CI, Bitbucket Pipelines, Woodpecker/Forgejo.
Reference
- CLI Reference — every command and flag.
- Troubleshooting — common issues and solutions.
- Licensing — free tier, commercial activation, and license management.