codebase-memory-mcp
A local tool that maps your code, docs and infrastructure into a graph so AI coding agents can query it.
- Stars
- 48k
- Forks
- 4k
- Licence
- MIT
- Language
- C
- Last update
- 4 days ago
- Type
- Dev tool
What it is
codebase-memory-mcp is a server that reads a code repository and builds a persistent local knowledge graph of its symbols, files, documentation and infrastructure files. It exposes this graph through the Model Context Protocol (MCP) so an AI coding agent can ask structural questions, such as what calls a function or what depends on a file, instead of reading many files. Indexing and queries run on your own machine, and the README says no hosted service or API key is needed.
- Ask an AI agent to find an unfamiliar feature and the code that implements it
- Trace a call path from one function to another across files
- Follow how a deployment configuration relates back to source
- Inspect the callers, dependencies and source snippets of a symbol from the CLI or an agent
Great for
- Developers who use an MCP-capable AI coding agent on a large codebase
- Engineers who want to find where a feature lives and how calls connect across files
- Teams that want to estimate which code and tests a change could affect
- People who prefer keeping source analysis on their own machine
Not ideal for
- Users who want a hosted service or cloud-managed indexing, since the README describes local processing only
- People without an MCP-capable agent or CLI experience, since setup assumes some terminal comfort
- Projects needing change-aware test selection for languages beyond the supported C/C++ harnesses, since Smart CI is described as experimental and limited to supported C/C++ test harnesses
How to get it
- 1
See the README
You'll need: curl and bash (macOS / Linux) · PowerShell (Windows)
Install the latest release (macOS / Linux)
curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bashStart CBM Lens after installation
codebase-memory-mcp daemon start --open --port=9749 - 2
Run it
- Run the install command for your platform from the README.
- Start CBM Lens with the daemon command if you want the visual interface.
- Choose the agent clients you use when the installer prompts, as it writes their configuration.
- Ask your agent a structural question about a repository you have indexed.
- 3
Get the most out of it
- Use the installer's agent setup to add skills, profiles and applicable hooks for your client.
- Use the integration matrix in the README to confirm which features your client supports.
- Share optional graph artifacts so a team can start from a common baseline.
- Try CBM Lens for source reading and call-path views if you prefer a visual approach.
Downloads
Strengths
- Runs locally with no hosted service or API key needed for indexing and queries, per the README
- Ships as a native binary; the README says no Docker or language runtime is required
- Integrates with many agent clients through an installer that writes their configuration
Trade-offs
- CBM Lens, the visual companion, is in open beta (v0.0.1), and Smart CI is labeled experimental
- Support varies by agent client, so you should check the integration matrix for yours
- The README is truncated, so details such as full language coverage and exact limits are not confirmed here
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| codebase-memory-mcp (this) | 48k | Dev tool | intermediate | 4 days ago | Developers who use an MCP-capable AI coding agent on a large codebase |
| Trail-Code | 11k | Dev tool | beginner | 4 days ago | Developers who use Claude Code, Cursor, Codex or similar agents on a medium or large repository |
| code-review-graph | 32.6k | Dev tool | beginner | 8 days ago | Developers who use an MCP-compatible AI coding assistant on a large repository |
| ECC | 295k | Dev tool | intermediate | 4 days ago | Developers who already use Claude Code and want a shared workflow for planning, testing and review |
| Understand-Anything | 89.7k | Dev tool | beginner | 11 days ago | Developers joining a new team who need to learn an unfamiliar codebase |
| agentic-awesome-skills | 49k | Dev tool | intermediate | 4 days ago | Developers using Codex or Claude who want agent-chosen instructions for a project |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.