Verified project record

DeusData/codebase-memory-mcp

deusdata/codebase-memory-mcp is a code intelligence MCP server that indexes local repositories into a persistent knowledge graph using tree-sitter AST analysis across 158 languages. It provides structural queries, semantic search, and cross-service linking to reduce token usage during codebase exploration.

Project overview

Project type: MCP · AI Coding · AI Search
Use cases: Knowledge Q&A · Coding & Development
Deployment: Refer to project documentation
License: MIT

Best for

Developers and AI engineers who use MCP-compatible coding agents and need to perform local structural and semantic queries on large codebases without transmitting source code externally.
Teams looking to share indexed codebase artifacts internally to reduce duplicate indexing overhead across local workstations.

Key capabilities

Indexes codebases into a persistent knowledge graph of functions, classes, call chains, HTTP routes, and cross-service links using tree-sitter AST analysis across 158 languages. Full-indexes an average repository in milliseconds and the Linux kernel (28M LOC) in 3 minutes.
Enhances tree-sitter parsing with lightweight C implementation of language type-resolution algorithms for Python, TypeScript/JavaScript/JSX/TSX, PHP, C#, Go, C, C++, Java, Kotlin, Rust, and Perl.
Provides vector search across the graph using bundled Nomic nomic-embed-code embeddings compiled into the binary, with 11-signal combined scoring including TF-IDF, RRI, AST profiles, and data flow.
Provides an interactive 3D UI at localhost:9749 for exploring the codebase knowledge graph as an optional UI binary variant.
Executes Cypher-like relationship traversal queries such as MATCH (f:Function)-[:CALLS]->(g) WHERE f.name = 'main' RETURN g.name against the knowledge graph.
Automatically detects and configures MCP entries, durable instructions, skills, and lifecycle hooks for 43 supported automatic/conditional client surfaces.
Optionally indexes new projects automatically on first connection and uses a background watcher for ongoing git-based change detection with configurable limits.
Allows every MCP tool to be invoked from the command line, supporting headless operations like indexing and querying.

Limitations and risks

The system does not include a built-in LLM and relies entirely on the connected MCP client for natural language processing.
Context hooks are withheld for Cursor because session injection races and readonly blocks MCP.
Child agents cannot use MCP for Cline, and automatic file hooks are withheld.
The tool reads the codebase and writes to agent configuration files, requiring user trust in the installer script.

Getting started

Run a one-line curl or PowerShell install script depending on your operating system.
The installer automatically detects and configures supported MCP client surfaces.
Restart your MCP-compatible agent after running the installer to load the new configuration.

Evidence and sources

GitHub project description: High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens.…
README: The fastest and most efficient code intelligence engine for AI coding agents. Full-indexes an average repository in milliseconds, the Linux kernel (28M LOC, 75K files) in 3 minute…
README: Plug and play — single static binary for macOS (arm64/amd64), Linux (arm64/amd64), and Windows (amd64). No Docker, no runtime dependencies, no API keys. Download → install → resta…
README: **One-line install** (macOS / Linux): ```bash curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash ``` With graph visualization UI: ```…
README: The install command auto-detects installed coding agents and configures their documented MCP entries plus durable instructions, skills, and lifecycle hooks where supported.

Reviewed at: 2026-07-17T09:35:42Z