Multi-agent AI Code Review

Code reviews that
actually learn

GHAGGA combines 17 static-analysis tools, multi-agent review, and project memory to review your Pull Requests with context that improves over time. You configure a provider chain: a gateway, a local CLI, or Ollama.

Install GitHub App Documentation Open Dashboard

17 Static Analysis Tools

Semgrep, Trivy, Gitleaks, Ruff, Biome, clippy, zizmor, and 10 more — security, linting, complexity, and license scanning. Zero LLM tokens consumed. SonarQube runs only when an MCP server is configured.

6 Review Modes

Simple, workflow, consensus, diagnostic, fan-out, and hybrid-4r. From one LLM pass to pinned multi-lens review.

Project Memory

Learns patterns, decisions, and past bugs across reviews. PostgreSQL full-text search with privacy stripping.

Delegated Runner

Local tools run on your repo's GitHub Actions runner. Public repos get unlimited Actions minutes. SonarQube stays off unless you configure MCP.

Get started in minutes

Zero to first review in 3 steps

1 Install Install the GitHub App on your repos
2 Configure Log in to the Dashboard, pick your LLM provider
3 Review Open a PR and get AI-powered review in ~1-2 min

Review Pipeline

From diff to insight in seconds

Layer 0 Static Analysis 17 tools · 7 always-on
Layer 1 Memory Search Past decisions · Patterns
Layer 2 AI Agents gateway · cli-bridge · ollama
Layer 3 Learn Extract · Persist
3,000+
Tests passing
4
Distribution modes
3
Provider modes
17
Analysis tools

Pricing

Free & Open Source

GHAGGA is free and open source (MIT license)
Reviews use a provider chain: gateway, a local CLI, or Ollama
Point the gateway at your model, or run a local CLI or Ollama
Static analysis runs on free GitHub Actions minutes