Local semantic code search engine for AI coding agents, reducing token use and costs by up to 50%.
lumen is worth checking the docs before setup with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
gh repo view ory/lumen --webOpen the official README and confirm the supported install method.
Add the server entry to your MCP client config.
Restart your agent and verify that the server tools appear.
Lumen is a tool that helps AI coding agents find relevant code quickly without reading entire files. It runs entirely on your computer using local embedding models and SQLite, so no data leaves your machine. By providing a semantic map of your codebase, it cuts token consumption and costs significantly while maintaining patch quality.
Lumen is a 100% local semantic code search engine designed for AI coding agents like Claude Code, Codex, and OpenCode. It indexes your codebase using Go AST and tree-sitter, then provides semantic search via local embedding models (Ollama or LM Studio) and SQLite-vec. No API keys, no cloud, no external database—just a single static binary and your own local embedding server. Benchmarks across 9 languages show cost reductions up to 39%, output token drops up to 66%, and session speedups up to 53%, all while maintaining patch quality. Lumen supports multiple agents and integrates as an MCP server or plugin. It is enterprise-ready, fully compliant, and works for both small and large codebases and monorepos.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 18 days ago.
207 GitHub stars indicate community interest.
19 open issues signal maintenance load.
NOASSERTION license detected.
Reduce Claude Code costs by providing semantic code context instead of reading entire files.
Speed up bug-fix sessions in large monorepos by quickly locating relevant code.
Enable offline, private code search for AI agents without sending data to the cloud.
Improve code generation quality by giving agents precise context from the codebase.
Requires running a local embedding server (Ollama or LM Studio) which consumes CPU/GPU resources.
Indexing large codebases may take significant time and disk space for the SQLite database.
207
Stars
23
Forks
19
Issues
NOASSERTION
License
Local codebase intelligence CLI and MCP server for AI coding agents with change-safety gates and audit evidence.
An offline MCP server that indexes your codebase for semantic search, code search, and git history retrieval.
Official MCP reference servers from Anthropic. Includes servers for filesystem, GitHub, Postgres, Slack, and more.
2 security/trust notes recorded.
Setup difficulty is 3/5.