codebase-memory-mcp

// Overview

codebase-memory-mcp is a code intelligence engine for AI coding agents. It parses a repository into a persistent knowledge graph of functions, classes, call chains, HTTP routes and cross-service links, then exposes that graph to an agent through 15 MCP tools — search, trace, architecture, impact analysis, index-coverage checks, Cypher queries, dead-code detection, cross-service HTTP linking and ADR management among them.

The point is to replace file-by-file exploration. When an agent has to answer "what calls this function" by grepping and reading, it burns a large fraction of its context on files it will discard. Querying a precomputed graph collapses that into a structured answer: the project reports five structural queries costing roughly 3,400 tokens where file-by-file search costs about 412,000, and its preprint (arXiv:2603.27277, evaluated across 31 real-world repositories) reports 83% answer quality with 10x fewer tokens and 2.1x fewer tool calls than file-by-file exploration.

Parsing is tree-sitter AST analysis across 162 languages, with grammars vendored and compiled into the binary so there is nothing to install per language. For Python, TypeScript/JavaScript/JSX/TSX, PHP, C#, Go, C, C++, Java, Kotlin, Rust and Perl, a Hybrid LSP layer adds semantic type resolution on top. It ships as a pure-C native executable for macOS (arm64/amd64), Linux (arm64/amd64) and Windows (amd64) — no Docker, no language runtime, no API key — and the installer configures detected agent clients across 45 supported surfaces. All processing is local; a built-in 3D graph visualisation is served from the binary itself at localhost:9749. Because the tool reads your codebase and writes to your agent configuration files, review the repository's SECURITY.md before running the install script.

// What it does

• Persistent knowledge graph of functions, classes, call chains, HTTP routes and cross-service links
• 15 MCP tools covering search, trace, architecture, impact analysis, Cypher queries, dead-code detection and ADR management
• 162 languages via vendored tree-sitter grammars, plus Hybrid LSP semantic type resolution for 12 of them
• Fast full indexing — the project reports the Linux kernel (28M LOC, 75K files) in 3 minutes and sub-millisecond structural queries
• Native executable for macOS, Linux and Windows with no Docker, language runtime or API key; processing stays on your machine
• Installer configures detected agent clients across 45 supported surfaces, and a built-in 3D graph UI runs at localhost:9749
• Infrastructure-as-code indexing: Dockerfiles, Kubernetes manifests and Kustomize overlays become graph nodes with cross-references

// Getting started

Installation is a single script that fetches the release binary and wires up your detected agent clients. The script reads your codebase and writes to agent configuration files by design — the full source is on GitHub if you would rather audit it first.

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

# 1. Download the installer
Invoke-WebRequest -Uri https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.ps1 -OutFile install.ps1

# 2. (Optional but recommended) Inspect the script
notepad install.ps1

# 3. Unblock the downloaded file
Unblock-File .\install.ps1

# 4. Run it
.\install.ps1

http://localhost:9749

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Cut an agent's token spend on a large repository by answering structural questions from a graph instead of reading files
• Run impact analysis before a refactor — trace every caller and cross-service HTTP link that a change would touch
• Give a coding agent a working map of an unfamiliar polyglot codebase without per-language tooling setup
• Find dead code, or query the graph directly with Cypher, across a repository of any supported language

// How codebase-memory-mcp compares

codebase-memory-mcp alongside other open-source code indexing & search tools AI/TLDR tracks, ranked by GitHub stars.