mcp-server-milvus Project Detailed Introduction

Project Overview

mcp-server-milvus is an open-source project developed by Zilliz Tech that provides a Model Context Protocol (MCP) server implementation for the Milvus vector database. This project enables LLM applications to seamlessly integrate and access the functionality of the Milvus vector database through a standardized protocol.

Core Features

Supported Operation Tools

The MCP server provides the following core tools:

• milvus_text_search: Searches documents using full-text search functionality.

  • Parameters: collection_name, query_text, limit, output_fields, drop_ratio

• milvus_vector_search: Performs vector similarity search on a collection.

  • Parameters: collection_name, vector, vector_field, limit, output_fields, metric_type

• milvus_query: Queries a collection using a filter expression.

  • Parameters: collection_name, filter_expr, output_fields, limit

• milvus_list_collections: Lists all collections in the database.

• milvus_create_collection: Creates a new collection with a specified schema.

  • Parameters: collection_name, collection_schema, index_params

• milvus_load_collection: Loads a collection into memory for searching and querying.

  • Parameters: collection_name, replica_number

• milvus_release_collection: Releases a collection from memory.

  • Parameters: collection_name

• milvus_insert_data: Inserts data into a collection.

  • Parameters: collection_name, data

• milvus_delete_entities: Deletes entities from a collection based on a filter expression.

  • Parameters: collection_name, filter_expr

System Requirements

• Python 3.10 or higher
• Running Milvus instance (local or remote)
• uv tool (recommended for running the server)

Installation and Configuration

Clone the Project

git clone https://github.com/zilliztech/mcp-server-milvus.git
cd mcp-server-milvus

Run the Server Directly

The recommended way is to use uv to run the server directly without installation:

uv run src/mcp_server_milvus/server.py --milvus-uri http://localhost:19530

Or by setting environment variables in the .env file:

uv run src/mcp_server_milvus/server.py

Supported Client Applications

The MCP server can be used with various LLM applications that support the Model Context Protocol:

Claude Desktop Integration

1. Install Claude Desktop
2. Configure the claude_desktop_config.json file:

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

Cursor Editor Integration

Cursor supports two configuration methods:

Method 1: Configure via GUI

1. Go to Cursor Settings > Features > MCP
2. Click the "+ Add New MCP Server" button
3. Fill in the form:
   • Type: Select stdio
   • Name: milvus
   • Command: /PATH/TO/uv --directory /path/to/mcp-server-milvus/src/mcp_server_milvus run server.py --milvus-uri http://127.0.0.1:19530

Method 2: Configuration File

Create a .cursor/mcp.json file:

{
  "mcpServers": {
    "milvus": {
      "command": "/PATH/TO/uv",
      "args": [
        "--directory",
        "/path/to/mcp-server-milvus/src/mcp_server_milvus",
        "run",
        "server.py",
        "--milvus-uri",
        "http://127.0.0.1:19530"
      ]
    }
  }
}

Environment Variable Configuration

• MILVUS_URI: Milvus server URI (alternative to the --milvus-uri parameter)
• MILVUS_TOKEN: Optional authentication token
• MILVUS_DB: Database name (defaults to "default")

Usage Examples

Usage in Claude Desktop

Users can interact with the Milvus database through natural language:

Query the list of collections:

What are the collections I have in my Milvus DB?

Search for documents:

Find documents in my text_collection that mention "machine learning"

Usage in Cursor

Create a new collection:

Create a new collection called 'articles' in Milvus with fields for title (string), content (string), and a vector field (128 dimensions)

Troubleshooting

Connection Issues

• Verify that the Milvus instance is running
• Check that the URI configuration is correct
• Ensure that firewall rules do not block the connection
• Try using 127.0.0.1 instead of localhost

Authentication Issues

• Verify that the MILVUS_TOKEN is correct
• Check if the Milvus instance requires authentication
• Ensure you have the correct permissions to perform the operation

Technical Architecture

The project is developed based on Python and adopts a modern development toolchain:

• Uses uv as a package management and running tool
• Follows the MCP standard protocol specification
• Provides complete Milvus database operation encapsulation
• Supports multiple AI application clients

Application Scenarios

This project is mainly suitable for the following scenarios:

• AI-driven code editor enhancement
• Context enhancement for chat interfaces
• Custom AI workflow construction
• Integration of vector databases and LLMs
• Semantic search and RAG (Retrieval Augmented Generation) applications

Through mcp-server-milvus, developers can easily integrate the powerful features of the Milvus vector database into various AI applications, providing users with a more intelligent and accurate interactive experience.