Lumen

orybyory
•
317
•

Enhance AI coding agents' efficiency by providing precise, local semantic search across large codebases, significantly reducing costs and latency.

Lumen is a 100% local semantic code search engine designed specifically for AI coding agents like Claude Code. It acts as a smart map for your codebase, allowing AI agents to find relevant functions, types, and modules by meaning rather than keyword matching. By integrating with local embedding models (Ollama or LM Studio) and using SQLite, Lumen keeps all data on your machine, ensuring privacy and compliance. This approach leads to measurable improvements, including 2-2.3x faster task completion and 63-81% cheaper API costs, without any subscriptions or external dependencies, making it ideal for teams working on large codebases.

Características Principales

0111 GitHub stars
02100% local semantic code search for AI agents
03Auto-indexing and incremental updates via Merkle tree change detection
04Support for 14 programming language families with semantic chunking
05Significant reduction in API costs and task completion time for AI agents
06Integrates with local embedding backends like Ollama and LM Studio

Casos de Uso

01Optimizing Claude Code's interaction with large and complex codebases
02Providing private, on-device semantic code search for sensitive projects
03Reducing operational costs and latency for AI-assisted coding tasks