A high-performance code intelligence MCP server that indexes codebases into a persistent knowledge graph with 158 languages and sub-ms queries.
- v0.12.1
- Published at
- 27d ago
- 46,740
- Stars
- 378,396
- Download
- Synced at
- 27d ago
Latest release
v0.12.1Oct 16, 2026, 11:35 PMDownloads slow? Switch the source anytime above
README
Project Overview
codebase-memory-mcp is a high-performance code intelligence MCP server that indexes codebases into a persistent knowledge graph. Average repo indexing in milliseconds, Linux kernel (28M LOC, 75K files) in 3 minutes.
Key Features
- 158 languages via tree-sitter AST analysis
- Hybrid LSP semantic type resolution
- Persistent knowledge graph of functions, classes, call chains, HTTP routes
- 15 MCP tools
- Zero dependencies, single static binary
- Built-in 3D graph visualization
- Infrastructure-as-code indexing
Target Audience
- AI coding agent developers
- Developers needing intelligent codebase understanding
- Users wanting to reduce context token usage
Quick Start
- Download the binary for your platform from GitHub Releases
- Run the
installcommand - Restart your AI agent
- Visit localhost:9749 for graph visualization
Core Advantages
- Extreme indexing speed: average repo in milliseconds
- 120x fewer tokens: ~3,400 tokens vs ~412,000
- 158 languages supported
- 100% local processing, code never leaves your machine
System Requirements
- OS: macOS, Linux, Windows
- Architectures: x86_64, arm64
- Additional Dependencies: None
Notes
- Reads codebases and writes to agent config files
- All processing happens 100% locally
Release history
Sort