TL;DR
An MCP server that indexes codebases into a knowledge graph, enabling AI coding agents to explore code quickly and efficiently.
Key features
Extreme indexing speed: Average repo in milliseconds, Linux kernel (28M LOC, 75K files) in 3 minutes
158 languages supported: tree-sitter AST analysis and Hybrid LSP semantic type resolution for 9 languages
Sub-millisecond queries: Structural queries for functions, classes, call chains, HTTP routes, cross-service links
14 MCP tools: Tools for various code intelligence tasks
Single static binary: Zero dependencies, runs on macOS, Linux, Windows
Local processing: All processing happens locally; code never leaves your machine
When to use it
Use with AI coding agents (supports 11 agents like Claude, Cursor, Copilot) to quickly understand and explore large codebases. Ideal when you want to reduce token usage by over 90% compared to file-by-file exploration while improving answer quality.