mcp-chromadb-repo-search

README

mcp-chromadb-repo-search

An MCP server for semantically searching GitHub repository content indexed in ChromaDB.

mcp-chromadb-repo-search turns a focused natural-language question into an embedding, queries repository chunks created by chromadb-repo-indexer, and returns concise Markdown excerpts with repository, file, line, and commit-aware source attribution. An optional reranking model can improve the ordering of retrieved candidates.

The server is read-only and supports local MCP over stdio and remote stateless MCP over Streamable HTTP.

Documentation

• Overview — purpose, architecture, and key features
• Installation — prerequisites, installation methods, and MCP client setup
• Usage — flags, environment variables, YAML configuration, transports, and runtime behavior
• MCP Tool Reference — search request parameters, response structure, and errors
• Examples — configurations, MCP calls, filters, reranking, and direct testing

Quick Start

Install the executable:

go install github.com/UnitVectorY-Labs/mcp-chromadb-repo-search@latest

Configure the required Chroma and embedding settings, then start the stdio server:

export CHROMA_REPO_SEARCH_SERVER_URL=https://chroma.example.com
export CHROMA_REPO_SEARCH_COLLECTION_NAME=repository-content
export CHROMA_REPO_SEARCH_EMBEDDING_API_URL=https://embeddings.example.com
export CHROMA_REPO_SEARCH_EMBEDDING_MODEL=your-embedding-model

mcp-chromadb-repo-search

The Chroma collection must first be populated by chromadb-repo-indexer, using the same embedding model expected by the collection. See Installation for a complete setup and Usage for all configuration options.

Development

go test ./...
go build ./...