cq27-dev/rag-rat

What it does

rag-rat is an MCP (Model Context Protocol) server that enables coding agents to understand and query entire codebases intelligently. It creates a local repository intelligence index combining hybrid BM25 full-text search with vector semantic search capabilities, allowing agents to deeply understand code relationships and structure.

Key features

• Hybrid BM25 + vector semantic search for code discovery
• Tree-sitter code graph analysis tracking callers, callees, and imports
• Multi-language support: Rust, TypeScript/TSX, Kotlin, C/C++, Python
• Optional compiler-grade SCIP oracle for enhanced code intelligence
• Git history queryable integration
• Works locally without external dependencies
• Cross-platform support (macOS, Windows, Linux)

How to set up

Clone the repository from GitHub, install dependencies, and configure as an MCP server in your AI tool's configuration. Detailed setup instructions are available in the project's GitHub repository. The tool integrates directly with compatible AI agents and development environments.