MCP.so
ログイン
サーバー

mcp-server-qdrant: A Qdrant MCP server

@qdrant

An official Qdrant Model Context Protocol (MCP) server implementation

概要

What is mcp-server-qdrant?

mcp-server-qdrant is an official Model Context Protocol server for Qdrant, the vector search engine. It acts as a semantic memory layer that enables LLM applications to store and retrieve information using vector embeddings. It is built for developers integrating AI tools with persistent, searchable memory.

How to use mcp-server-qdrant?

Configure the server via environment variables (e.g., QDRANT_URL, COLLECTION_NAME, EMBEDDING_MODEL). Run it with uvx mcp-server-qdrant or via Docker. The server exposes two tools: qdrant-store to store information and qdrant-find to retrieve relevant information from Qdrant. It supports stdio, sse, and streamable-http transports.

Key features of mcp-server-qdrant

  • Semantic memory layer on top of Qdrant vector search.
  • Store and retrieve information with optional metadata.
  • Automatic collection creation if it does not exist.
  • Configurable embedding model (FastEmbed models only).
  • Works with any MCP-compatible client (Claude Desktop, Cursor, VS Code, etc.).
  • Customizable tool descriptions via environment variables.

Use cases of mcp-server-qdrant

  • Give LLM applications persistent, searchable memory for conversations or knowledge.
  • Store and retrieve code snippets with semantic search in Cursor or Windsurf.
  • Enable Claude Code to search over existing codebases semantically.
  • Build a personal knowledge base that AI assistants can query.
  • Share a remote semantic memory server across a team using SSE transport.

FAQ from mcp-server-qdrant

What are the main tools provided?

The server provides qdrant-store to store information (with optional JSON metadata) and qdrant-find to retrieve relevant information based on a query.

What embedding models are supported?

Currently only FastEmbed models are supported. The default model is sentence-transformers/all-MiniLM-L6-v2.

Can I run the server locally without a remote Qdrant instance?

Yes. Set QDRANT_LOCAL_PATH to a local database path; do not set QDRANT_URL at the same time.

Which transport protocols are available?

The server supports stdio (default), sse, and streamable-http. Use the --transport flag to select one.

Does the server require an API key for Qdrant?

It is optional. You can set QDRANT_API_KEY if your Qdrant server requires authentication.

タグ

「データベース」の他のコンテンツ