# mcp-server-qdrant

An official Model Context Protocol server that lets AI agents store and retrieve memories in a Qdrant vector database, acting as a semantic memory layer for LLM applications.

mcp-server-qdrant is an official MCP server for Qdrant, a vector search engine. It exposes two tools to any MCP client: `qdrant-store`, which saves a piece of information and optional metadata into a named collection, and `qdrant-find`, which retrieves information relevant to a query. This lets an LLM application keep and search a semantic memory layer backed by Qdrant.

Configuration is done through environment variables, including the Qdrant server URL or a local database path, the API key, the default collection name, and the embedding provider and model, FastEmbed by default. The only command-line argument is `--transport`, which selects the transport protocol.

Website: https://github.com/qdrant/mcp-server-qdrant