Codebase RAG

Self-hosted semantic code search over local codebases using ChromaDB and sentence-transformer embeddings with hybrid keyword reranking.

A self-hosted code search server that indexes local codebases using vector embeddings stored in ChromaDB. Supports Python AST-based chunking for Python files and character-based chunking for other languages, with hybrid search combining vector similarity and keyword reranking for improved relevance. Provides tools for semantic code search, project reindexing, file listing, and content retrieval across multi-project setups. Includes automation scripts for auto-discovery of Git repositories, file-watching with debounced reindexing, and post-commit Git hooks for incremental updates.