Introduction

What is Reviewate?

Reviewate is an open-source AI agent that automatically reviews your pull requests. It uses a multi-agent pipeline to analyze code changes, explore your codebase for context, and post actionable feedback directly on PRs.

Unlike traditional AI code reviewers that just scan the diff and hallucinate issues, Reviewate takes a codebase-aware approach: review agents use search tools to explore your codebase and verify their findings, then a separate fact-checker verifies every claim before posting.

How You Can Use It

Reviewate offers three usage modes depending on your needs:

• Local CLI — Test on any PR right now. Great for evaluation.
• CI Pipeline — Add to GitHub Actions or GitLab CI. No server required.
• Self-Hosted Platform — Full dashboard with webhook integration, team management, and centralized config. Deploy with Docker Compose.

Why Reviewate?

• Codebase-aware review — Agents explore your codebase with search tools to verify every finding against actual code
• Fact-checked — Every claim is verified against actual code before posting
• Context-aware — Links issues, understands your architecture, removes duplicate comments
• Powered by Claude — Uses Anthropic's Claude models via the Claude Agent SDK
• Self-hosted — Your code never leaves your infrastructure
• Isolated execution — Each review runs in its own container (Docker or Kubernetes)