Codebase Memory MCP
A knowledge graph of your codebase, in one static binary
What it is
An MCP server that indexes a repository into a persistent knowledge graph: tree-sitter parsing across 158 languages, deeper type resolution for 12, sub-millisecond graph queries, dead-code detection, and architecture analysis — shipped as a single zero-dependency binary.
Why it's interesting
Attacks the sorest point in agentic coding — context efficiency — with a local-first design that claims ~99% token savings over raw file reads, backed by an arXiv preprint and unusually serious supply-chain hygiene (SLSA L3, Sigstore-signed releases).
Use cases
- Giving Claude Code or any MCP client structural memory of a large repo
- Impact analysis and dead-code detection
- Fast code navigation for agents without API calls
Who it's for
Developers running MCP-capable agents on medium-to-large codebases
Setup
Easy. Download one binary, register it as an MCP server in your client config
Limitations & cautions
Semantic resolution covers 12 languages (the rest are syntax-only), and the graph is only as fresh as your last re-index. Binary-first distribution — verify provenance via the provided attestations.
Editorial takeaway
If your agent burns most of its budget re-reading files, this is the cure — and its release engineering should embarrass much bigger projects.