Qdrant

mcp-server-qdrant is an official Model Context Protocol server for the Qdrant vector search engine. It lets an LLM application keep and retrieve memories, acting as a semantic memory layer on top of a Qdrant database.

The server exposes two tools. qdrant-store saves a piece of information, with optional metadata, into a collection. qdrant-find takes a natural language query and returns the stored information most relevant to it, up to a configurable limit that defaults to 10 results. Setting QDRANT_READ_ONLY disables the store tool for read-only use.

It is aimed at developers who build AI-powered IDEs, chat interfaces or custom AI workflows and want a standard way to connect their models to stored context. The page includes setup steps for Claude Desktop, Claude Code, Cursor, Windsurf and VS Code, along with a Docker image.

The server runs over stdio by default, or over SSE or streamable HTTP for remote clients. It can connect to a Qdrant server by URL with an API key, or to a local Qdrant database path. Embeddings are generated with FastEmbed models, with sentence-transformers/all-MiniLM-L6-v2 as the default. A collection is created automatically if it does not already exist.

The project is licensed under Apache-2.0 and the repository is public. The page does not list prices for the server itself. Running it still requires a Qdrant database, either self-hosted or hosted by Qdrant, and the page does not say what that costs.