Milvus

by Zilliz

v0.1.1.dev9

Repository Docs

Databases & Data

vector-databasemilvussearchrag

Summary

Search and manage Milvus vector collections from an AI client — vector, full-text, and hybrid queries.

Features

Vector, full-text, and hybrid search over Milvus collections
Collection and index management
Works with Milvus Lite, self-hosted, and Zilliz Cloud

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

Open your agent's MCP configuration file.
Merge the snippet below into it.
Restart the agent — the "Milvus" tools will be available.

{
  "mcpServers": {
    "milvus": {
      "command": "uvx",
      "args": [
        "mcp-server-milvus",
        "--milvus-uri",
        "http://localhost:19530"
      ]
    }
  }
}

Claude Code

Add the server with a single CLI command.

claude mcp add milvus -- uvx mcp-server-milvus --milvus-uri http://localhost:19530

Claude Desktop

Add the server to claude_desktop_config.json (Settings → Developer → Edit Config).

{
  "mcpServers": {
    "milvus": {
      "command": "uvx",
      "args": [
        "mcp-server-milvus",
        "--milvus-uri",
        "http://localhost:19530"
      ]
    }
  }
}

Cursor

{
  "command": "uvx",
  "args": [
    "mcp-server-milvus",
    "--milvus-uri",
    "http://localhost:19530"
  ]
}

Windsurf

{
  "command": "uvx",
  "args": [
    "mcp-server-milvus",
    "--milvus-uri",
    "http://localhost:19530"
  ]
}

VS Code

{
  "command": "uvx",
  "args": [
    "mcp-server-milvus",
    "--milvus-uri",
    "http://localhost:19530"
  ]
}

Description

The Milvus MCP server from Zilliz gives an assistant access to Milvus, one of the most widely deployed open-source vector databases.

Capabilities

Search — dense vector similarity, sparse full-text, and hybrid queries combining both, with metadata filtering
Collection management — create, describe, load, release, and drop collections
Data operations — insert, upsert, delete, and query by expression
Index management for tuning the accuracy/speed trade-off

Deployment options

The same server works against Milvus Lite (an embedded file-based mode ideal for local development), a self-hosted Milvus cluster, and Zilliz Cloud. That range means the same configuration pattern carries from a laptop prototype to production.

Where it fits

Useful both as a retrieval backend an assistant can query during a task, and as a development aid — inspecting what is actually in a collection is the fastest way to debug a RAG pipeline returning nonsense.

Runs with uvx mcp-server-milvus --milvus-uri <uri>.