MCP Server for Milvus

Name: MCP Server for Milvus
Author: zilliztech

@zilliztech

代码仓库

关于 MCP Server for Milvus

Model Context Protocol Servers for Milvus

#mcp #milvus #llm-integration

基本信息

分类

数据库

许可证

Apache-2.0

运行时

python

传输方式

stdio

发布者

zilliztech

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "mcp-server-milvus": {
      "command": "uv",
      "args": [
        "run",
        "src/mcp_server_milvus/server.py",
        "--milvus-uri",
        "http://localhost:19530"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is MCP Server for Milvus?

The MCP Server for Milvus is a Model Context Protocol (MCP) server that provides access to Milvus vector database functionality. It enables LLM applications like Claude Desktop, Cursor, and custom MCP clients to interact with Milvus for vector search, text search, and collection management.

How to use MCP Server for Milvus?

Run the server with uv and a running Milvus instance: uv run src/mcp_server_milvus/server.py --milvus-uri http://localhost:19530. Alternatively, set environment variables in a .env file in src/mcp_server_milvus/. The server supports three communication modes: stdio (default), SSE, and Streamable HTTP, selectable via command-line flags.

Key features of MCP Server for Milvus

Provides full-text, vector, hybrid, and similarity search tools
Allows querying collections with filter expressions
Supports creating collections with quick or custom schema
Lists all collections in the database
Works with Claude Desktop and Cursor via MCP
Offers stdio, SSE, and Streamable HTTP transport modes

Use cases of MCP Server for Milvus

AI-powered code editors performing semantic code search in a vector database
Chat interfaces retrieving relevant documents via hybrid search
Custom AI workflows that query and manage vector collections
Automated collection creation and schema setup for new projects

FAQ from MCP Server for Milvus

What are the transport modes and how do they differ?

Stdio mode communicates over standard input/output (default). SSE mode uses HTTP Server-Sent Events for multiple clients. Streamable HTTP mode provides HTTP with streaming support, recommended for production, and optionally supports stateless operation.

What prerequisites are required to run this server?

Python 3.10 or higher, a running Milvus instance (local or remote), and uv (recommended) installed. No additional server dependencies are required beyond those.

What Milvus version is required?

The server works with any standard Milvus instance. Note that the milvus_text_similarity_search tool requires Milvus 2.6.0 or above and an embedding function configured on the server.