LanceDB MCP Server

Example Use Cases

• 1

  Persist and retrieve dense vector embeddings with associated text metadata for downstream reasoning or retrieval.

• 2

  Perform similarity search (nearest-neighbor retrieval) over a local vector store with configurable dimensions and result limits.

• 3

  A simple MCP-compatible server to integrate LanceDB as a resource within multi-agent or desktop assistant setups (e.g., Claude Desktop).

Description

This server implements a Model Context Protocol (MCP) interface for LanceDB, allowing agents to create and manage vector tables, insert vectors with associated text metadata, and run similarity searches. It is designed to be run locally (configurable storage path) and integrated into agent workflows such as Claude Desktop. The project is open-source, focused on efficient vector storage and retrieval for embedding-based applications.