Qdrant
FreeQdrant 官方集成，基于 Qdrant 向量搜索引擎实现语义记忆层。
FreeFree tier
Inputs: textOutputs: text
About Qdrant
An official MCP (Model Context Protocol) server implementation for Qdrant vector search engine, providing a semantic memory layer for LLM applications. It enables AI systems to persistently store and retrieve contextual information using vector embeddings. The server exposes two primary tools: qdrant-store for storing information with optional metadata, and qdrant-find for performing semantic searches. Configuration is handled via environment variables, supporting local or remote Qdrant instances, customizable embedding models (default: sentence-transformers/all-MiniLM-L6-v2 via FastEmbed), read-only mode, and integration with FastMCP for easy deployment.
Key Features
Acts as a semantic memory layer on top of Qdrant vector search engine
Includes qdrant-store (store information with optional metadata) and qdrant-find (semantic search) tools
Supports configurable embedding models via FastEmbed (default: all-MiniLM-L6-v2)
Local and remote Qdrant database support (QDRANT_URL or QDRANT_LOCAL_PATH)
Read-only mode option (disables store tool)
Built on FastMCP for easy integration with MCP-compatible clients
Pros & Cons
Pros
- Official Qdrant integration ensures compatibility and performance
- Provides a semantic memory layer for LLMs, enabling long-term context
- Open-source and free to use
- Easy to configure via environment variables with sensible defaults
- Supports both local and cloud Qdrant deployments
Cons
- Requires a running Qdrant instance (local or remote) or local database setup
- Depends on embedding model availability and performance
- Currently only supports FastEmbed as the embedding provider
Best For
AI-powered IDE integration with persistent memoryChat interface enhancement (conversation history, user context)Custom AI workflows requiring vector storage and retrievalSemantic search and context retention for LLM applications
FAQ
What is the MCP server for Qdrant?
It is an official Model Context Protocol server that enables LLM applications to store and retrieve memories using Qdrant's vector search engine, acting as a semantic memory layer.
How do I configure the server?
Configuration is done via environment variables, including QDRANT_URL, QDRANT_API_KEY, COLLECTION_NAME, QDRANT_LOCAL_PATH, EMBEDDING_PROVIDER, and EMBEDDING_MODEL.
Can I run the server in read-only mode?
Yes, by setting QDRANT_READ_ONLY=true, the qdrant-store tool is disabled, allowing only retrieval operations.