Codebase-memory MCP server indexes repos in milliseconds with 99% fewer tokens
DeusData's `codebase-memory-mcp` is a high-performance MCP server that indexes codebases into a persistent knowledge graph, supporting 158 languages, sub-millisecond queries, and claiming 99% fewer tokens — shipped as a single static binary with zero dependencies.
Score breakdown
The server's persistent knowledge graph approach reduces the token cost of codebase exploration by a claimed 99%, directly addressing one of the primary bottlenecks for AI coding agents working on large repositories.
- 01Indexes an average repository in milliseconds; the Linux kernel (28M LOC, 75K files) in 3 minutes
- 02Answers structural queries in under 1ms and claims 99% fewer tokens
- 03Supports 158 languages via tree-sitter AST analysis
DeusData's `codebase-memory-mcp` is an open-source MCP server designed to give AI coding agents fast, structured access to large codebases. It builds a persistent knowledge graph of functions, classes, call chains, HTTP routes, and cross-service links by parsing source code through tree-sitter AST analysis. The project claims to index an average repository in milliseconds and benchmarks the Linux kernel — 28 million lines of code across 75,000 files — in 3 minutes, with structural queries returning in under 1ms. It supports 158 languages and claims 99% fewer tokens compared to alternative approaches.
Beyond basic tree-sitter parsing, the server adds Hybrid LSP semantic type resolution for a subset of languages including Python, TypeScript, JavaScript, JSX, TSX, PHP, C#, Go, C, C++, Java, Kotlin, and Rust.
Beyond basic tree-sitter parsing, the server adds Hybrid LSP semantic type resolution for a subset of languages including Python, TypeScript, JavaScript, JSX, TSX, PHP, C#, Go, C, C++, Java, Kotlin, and Rust. It exposes 14 MCP tools and is described as plug-and-play across 11 coding agents. The project ships as a single static binary for macOS, Linux, and Windows with zero dependencies, installable via a provided install script. The underlying design and benchmarks are described in the preprint *Codebase-Memory: Tree-Sitter-Based Knowledge Graphs for LLM Code Exploration via MCP* (arXiv:2603.27277). The repository has accumulated 6.3k stars and 521 forks on GitHub.
Key facts
- 01Indexes an average repository in milliseconds; the Linux kernel (28M LOC, 75K files) in 3 minutes
- 02Answers structural queries in under 1ms and claims 99% fewer tokens
- 03Supports 158 languages via tree-sitter AST analysis
- 04Hybrid LSP semantic type resolution for Python, TypeScript/JavaScript/JSX/TSX, PHP, C#, Go, C, C++, Java, Kotlin, and Rust
- 05Ships as a single static binary for macOS, Linux, and Windows with zero dependencies
- 06Exposes 14 MCP tools and is described as compatible with 11 coding agents
- 07Design and benchmarks documented in arXiv preprint arXiv:2603.27277
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 18, 2026 · 10:40 UTC. How this works →