Codebase Memory MCP

by Martin VogelConnectorWrite code

Gives your coding agent a local, searchable map of your code, docs and infrastructure

About Codebase Memory MCP

Codebase Memory MCP connects your coding agent to a local knowledge graph of a project's code, docs and infrastructure, so it can search, trace calls and estimate the reach of a change.

For DevelopersA team sharing a graph baseline

Made by Martin Vogel

Martin Vogel, who runs the DeusData GitHub account, publishes it as an MIT-licensed project with releases for macOS, Linux and Windows.

Stars: ★ 48k
License: MIT
Cost: Free
Level: Intermediate
Runs: Local

Screenshots

From the project's own repository · 3 images, swipe or use the arrows.

CBM Lens Explore view showing limits.c in the source reader beside its file impact summary and graph dependencies — Image: Martin Vogel (MIT). Original file

CBM Lens architecture map of this repository's folders as nested platforms with arrows for relationships — Image: Martin Vogel (MIT). Original file

Overview

Codebase Memory MCP indexes a project into a persistent knowledge graph on your computer and offers 17 MCP tools, so your coding agent can find a feature, follow a call across services or see what a change could affect without re-reading every file. It covers code, documentation and configuration such as Docker, Kubernetes and Terraform. A one-line installer sets up the coding agents it finds, and an optional browser view called CBM Lens shows the graph. It suits developers exploring an unfamiliar repository.

Indexing and queries run locally with no hosted service or API key, which can cut repeated file reads.

Try these first

Start with the first one: it is a small task you can check easily.

Index this project.First try · writes a local index
Who calls [function name], and what does it call?read-only
Which parts of the project could this change affect: [describe the change]?read-only
Find where [feature] is implemented and list the main files.read-only

Example prompts written by Agent Flows Hub.

Capabilities

From the creator's documentation.

Toolset	What your agent can do	Access
Code search	Finds symbols, text and semantic matches with callers, dependencies and snippets	Read
Change impact	Shows which symbols and tests a change could affect	Read
Docs and infrastructure	Links documentation and Docker, Kubernetes, Terraform and CI files to code	Read
Architecture decisions	Reads and edits the project's architecture decision record stored with the index	Read + write
Agent setup	The installer writes client config, skills and hooks for the agents you select	Read + write
CBM Lens	Optional local browser view of the graph; a chat model downloads only if you enable it	Read, one optional download

Good to know

You open an unfamiliar repository and need to find where a feature lives or what a change touches. The server builds a local graph of code, docs and configuration, and your agent queries it instead of reading file after file.

Limits

The installer writes MCP configuration, skills and, by default, hooks for the coding agents it detects; --no-hooks removes the hooks and keeps the MCP setup.
CBM Lens, the browser view, is an open beta (v0.0.1), and its optional local chat model downloads about 567 MB only after you enable it.

Notes by Agent Flows Hub (original project name: codebase-memory-mcp), written from the creator's documentation, not from running it ourselves. codebase-memory-mcp README: quick start and tools · codebase-memory-mcp documentation

Stuck setting it up?

Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.

Get help

How to start

Open your AI agentClaude Code, Codex, Cursor or another agent that supports MCP.
Paste the install requestIt asks your agent to read the official instructions, tell you what it will connect to and wait for your OK.
Give it a first taskInstall it, restart your coding agent, say "Index this project" on a small repository, then ask who calls one function and open that function to compare.

Steps follow the creator's instructions. The request tells your agent to read them first and wait for your OK.

Before you start

An AI agent that supports MCP, such as Claude Code, Codex or Cursor.
A coding agent that supports MCP, such as Claude Code or Codex
A terminal to run the one-line installer
Free. Nothing to pay.

Details

Made by: Martin Vogel
License: MIT Terms
Works with: Claude Code, Codex, Cursor, Gemini CLI, OpenCodeAs stated by the creator
Runs on: Your computer

View source Official docs Report a problem

For AI agents

This page as a Markdown install brief:

/en/library/codebase-memory-mcp/install.md

Other connectors

DaVinci Resolve MCP Lets your AI agent sort media, build timelines, grade and render in DaVinci Resolve
Desktop Commander MCP Lets your AI app read, edit and organize files and run commands on your computer
earshot Lets your coding agent hear audio and judge sound effects, music and film soundtracks
Exa MCP Server Lets your AI agent search the web and read pages as clean text, without an account
Figma MCP Server Lets your AI agent turn Figma designs into code and create designs in your Figma files
Firecrawl MCP Server Lets your AI agent search the web and turn pages and documents into clean, readable text

See all 57 connectors · An MCP server is missing or disconnected

GitHub