Codebase-Memory-MCP: High-Performance Code Intelligence for AI Agents

Introduction

Codebase-Memory-MCP is an exceptionally fast and efficient code intelligence engine tailored for AI coding agents. It transforms your codebase into a persistent knowledge graph, indexing an average repository in milliseconds and even the Linux kernel in just 3 minutes. Supporting 158 languages through Tree-sitter and Hybrid LSP, it enables sub-millisecond structural queries and drastically reduces token consumption for AI interactions. Packaged as a single, static binary with zero dependencies, it offers plug-and-play integration across 11 popular coding agents.

Installation

Getting started with Codebase-Memory-MCP is straightforward.

For macOS / Linux, use the following one-line command:

curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash

To include the built-in graph visualization UI, use:

curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash -s -- --ui

For Windows (PowerShell), download and run the installer:

Invoke-WebRequest -Uri https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.ps1 -OutFile install.ps1
.\install.ps1

After installation, restart your coding agent. You can then instruct your agent with "Index this project" to begin. To keep your installation up to date, simply run:

codebase-memory-mcp update

Examples

Codebase-Memory-MCP provides a powerful command-line interface (CLI) for direct interaction. Here are some examples:

• Index a repository:

  codebase-memory-mcp cli index_repository '{"repo_path": "/path/to/your/repo"}'
  

• Search for functions matching a pattern:

  codebase-memory-mcp cli search_graph '{"name_pattern": ".*Handler.*", "label": "Function"}'
  

• Trace inbound calls to a specific function:

  codebase-memory-mcp cli trace_path '{"function_name": "ProcessOrder", "direction": "inbound"}'
  

• Execute a Cypher-like graph query:

  codebase-memory-mcp cli query_graph '{"query": "MATCH (f:Function) RETURN f.name LIMIT 5"}'
  

Additionally, if you installed the UI variant, you can explore your codebase's knowledge graph visually by running codebase-memory-mcp --ui=true --port=9749 and opening http://localhost:9749 in your browser.

Why Use Codebase-Memory-MCP

Codebase-Memory-MCP stands out with several compelling features:

• Extreme Indexing Speed: Indexes the Linux kernel (28M LOC) in 3 minutes and an average repository in milliseconds, thanks to a RAM-first pipeline.
• Broad Language Support with Hybrid LSP: Parses 158 languages using vendored Tree-sitter grammars, enhanced by a lightweight C implementation of language type-resolution algorithms (Hybrid LSP) for 10+ languages, providing semantic accuracy without external language servers.
• Exceptional Token Efficiency: Achieves up to a 99.2% reduction in tokens for AI agents, replacing dozens of grep/read cycles with single graph queries.
• Plug and Play Simplicity: Delivered as a single static binary for macOS, Linux, and Windows, requiring zero runtime dependencies or API keys.
• Seamless Multi-Agent Integration: Automatically configures MCP entries, instruction files, and pre-tool hooks for 11 popular coding agents, including Claude Code, Codex CLI, and Gemini CLI.
• Team-Shared Graph Artifacts: Supports committing a compressed knowledge graph snapshot to your repository, allowing teammates to skip full re-indexing and accelerate onboarding.
• Comprehensive Security: Ensures trust with SLSA Level 3 cryptographic build provenance, VirusTotal scans on all binaries, Sigstore keyless signatures, and zero runtime dependencies.
• Powerful MCP Tools: Offers 14 specialized tools for architecture overview, semantic search, call graph analysis, dead code detection, cross-service linking, and Cypher-like queries.

Links

• GitHub Repository: DeusData/codebase-memory-mcp
• Research Paper (arXiv): Codebase-Memory: Tree-Sitter-Based Knowledge Graphs for LLM Code Exploration via MCP