Codebase Memory MCP is an open-source MCP server by Deusdata, designed to enhance AI coding agents. It operates by indexing code repositories into a persistent knowledge graph, providing structural awareness without external dependencies. This free tool runs entirely on the user's machine, ensuring zero telemetry and local data storage in SQLite. It integrates with various AI clients to optimize code exploration, reduce token usage, and assist with impact analysis, making it a valuable addition for developers managing complex codebases.

What does Codebase Memory MCP connect to?

Codebase Memory MCP connects directly to AI coding agents, functioning as an MCP client server. It integrates with popular AI clients such as Claude Code, Codex CLI, Gemini CLI, Cursor, Zed, Aider, VS Code, and OpenCode. The tool achieves this by indexing code repositories into a persistent knowledge graph, which serves as a local data source for AI agents. All program state, including the indexed code data, is stored locally within a SQLite database on the user's machine.

Core features and tools

The tool exposes multiple callable tools, with counts cited between 11 and 15. Key capabilities include search_graph for navigating code structures, trace_path to follow execution flows, and detect_changes for code modification analysis. It also offers query_graph for Cypher-style queries, trace_call_path, get_architecture for system overviews, dead-code detection, semantic_query, snippets extraction, architecture summaries, and index_repository to build the knowledge graph.

Who is it for?

Codebase Memory MCP is primarily for developers who work with AI coding agents, particularly those managing large and intricate code repositories. It helps agents quickly understand codebase structure, identify dependencies, and analyze changes. The tool is ideal for enhancing the efficiency and accuracy of AI-assisted development tasks. Skip this if you do not use AI coding agents or primarily work with very small, simple projects.

How to get started

No credentials are required. It communicates via local process execution.

Use cases

• Structural code exploration and understanding: Enables AI agents to query a knowledge graph for functions, classes, call chains, HTTP routes, and cross-service links, providing instant structural awareness.
• Reducing token usage and improving query speed: Allows agents to query the pre-built graph instead of repeatedly scanning raw files, significantly reducing token consumption and accelerating responses to structural questions.
• Impact analysis and dead code detection: Can be used to determine the blast radius of a code change or identify unused functions within a codebase.

Codebase Memory MCP offers significant advantages for developers employing AI coding agents, providing a local, private, and efficient method for code intelligence. Its ability to index large repositories into a queryable knowledge graph reduces AI agent overhead and improves response quality. The local-first design ensures data privacy and ease of setup.