AI Review
AI Review is a full-stack platform that automatically reviews your merge requests and pull requests using AI models. It integrates directly with GitLab and GitHub via webhooks, runs isolated review jobs on distributed runners, and posts structured feedback back to your code platform.Quickstart
Get AI Review running and your first automated review in minutes
Self-Hosting
Deploy AI Review on your own infrastructure with Docker
Platform Integration
Connect GitLab or GitHub and configure webhooks
API Reference
Explore the full REST API surface
What AI Review does
When a developer opens a merge request (GitLab) or pull request (GitHub), AI Review:- Receives the webhook event from the platform
- Checks whether the project has automated review enabled and has a valid AI configuration
- Enqueues the review task via BullMQ
- A Runner picks up the task, executes the AI review in an isolated Docker environment
- The result — inline comments, a summary, a rating, and per-file analysis — is posted back to the platform and stored in the database
System components
Server
Hono API service. Handles webhooks, orchestrates reviews, manages platform integrations, and serves the frontend in production.
Web
React 19 admin dashboard. Provides project management, review history, runner status, trends, notifications, and AI assistant.
Runner
Standalone executor. Polls the server for tasks, runs AI review in an isolated Docker container, and reports results back.
Key features
- Automated reviews — Trigger on GitLab MR open/update/reopen and GitHub PR opened/synchronize/reopened
- Distributed execution — Runners execute reviews in isolated Docker containers, separate from the main service
- Configurable AI — Templates, rules, and rating systems per project
- Rich results — Per-file comments, review rounds, ratings, and feedback stored and displayed in the dashboard
- Multi-platform — Supports GitLab (self-hosted or cloud) and GitHub (cloud or enterprise)
- Notifications — Email, Slack, DingTalk, Feishu, WeChat, and browser push
- AI assistant — Chat interface for ad-hoc AI queries with session management
- Observability — Health checks, metrics, structured logs, backups
Technology stack
| Layer | Technology |
|---|---|
| API Server | Hono, Node.js 18+ |
| Database | PostgreSQL + Drizzle ORM |
| Queue | BullMQ + Redis |
| Frontend | React 19, TanStack Router, TanStack Query, Vite |
| Auth | Better Auth (session + API key) |
| AI | Volcengine (current active provider) |
| Deployment | Docker, Docker Compose |
Next steps
Quickstart
Run AI Review locally in under 5 minutes
Architecture
Understand how the components fit together