Milvus
FreeZilliz/Milvus 官方集成，搜索、查询和交互 Milvus 向量数据库中的数据。
FreeFree tier
About Milvus
MCP Server for Milvus is an open-source server that implements the Model Context Protocol (MCP) to enable seamless integration between LLM applications and Milvus vector databases. It provides a standardized way to search, query, and interact with vector data, making it ideal for AI-powered IDEs, chat interfaces, and custom AI workflows. The server supports multiple communication modes (stdio, SSE, and Streamable HTTP), can be run directly with uv, and is configurable via environment variables or command-line arguments.
Key Features
Implements Model Context Protocol (MCP) for standardized LLM integration
Supports stdio, SSE, and Streamable HTTP communication modes
Runs directly with uv – no installation required
Configurable via environment variables or command-line arguments
Prerequisites: Python 3.10+ and a running Milvus instance
Built and maintained by Zilliz (zilliztech)
Pros & Cons
Pros
- Open source and completely free to use
- Easy to deploy with uv – no complex setup
- Multiple communication modes for flexibility (stdio, SSE, HTTP)
- Official integration from Zilliz, ensuring compatibility with Milvus
- Designed for production with Streamable HTTP support
- Works seamlessly with common MCP clients like Claude Desktop and Cursor
Cons
- Requires a running Milvus instance (local or remote) – adds dependency
- Limited documentation beyond the README and GitHub repository
- Primarily targets Python 3.10+ environments
- No built-in authentication or authorization mechanisms (relies on Milvus itself)
Best For
AI-powered IDE integration (e.g., Claude Desktop, Cursor)Enhancing chat interfaces with vector database contextCreating custom AI workflows that leverage vector searchBuilding retrieval-augmented generation (RAG) pipelinesQuerying and managing Milvus collections via LLM agents
FAQ
What are the prerequisites for using this MCP server?
You need Python 3.10 or higher, a running Milvus instance (local or remote), and uv installed (recommended).
How do I run the MCP server?
Clone the repository, then run: uv run src/mcp_server_milvus/server.py --milvus-uri http://localhost:19530. You can also configure via a .env file.
What communication modes does the server support?
It supports stdio (default), SSE (Server-Sent Events), and Streamable HTTP modes. Use --sse or --streamable-http flags to switch.
Is this an official MCP server from Zilliz?
Yes, the repository is hosted under zilliztech on GitHub, indicating it is the official MCP integration for Milvus.