AI-Powered Code Review Server
Built a code review server that uses LLMs via OpenRouter with MCP Protocol support for direct integration with Claude Code
TypeScriptExpressReactViteZodDockerMCP Protocol
Completed January 2026
AI-Powered Code Review Server
Built a code review server that uses LLMs via OpenRouter to analyze code and provide feedback with issues and risk scores.
Features
- React dashboard where you can paste code or diffs and get feedback
- Issues detection with severity levels
- Risk scoring for code changes
- MCP Protocol support for direct integration with Claude Code
- Docker containerization for easy deployment
Technical Stack
- Frontend: React + Vite for a fast, modern UI
- Backend: Express.js with TypeScript
- Validation: Zod for runtime type checking
- AI Integration: OpenRouter for LLM access
- Protocol: MCP (Model Context Protocol) for editor integration
- Deployment: Docker and Docker Compose
Architecture
The server exposes a RESTful API for code review requests and also implements the MCP protocol for seamless integration with AI-powered editors like Claude Code.
Lessons Learned
This project taught me about integrating LLMs into development workflows and the importance of providing structured, actionable feedback to developers.