mcp-registry/mcp-server-milvus

    ==================
      
       /// MCP ///
      /// MCP ///
        
    ==================
        
    [server:online]
    [protocol:ready]

mcp-server-milvus

by zilliztech

Model Context Protocol (MCP) server that exposes Milvus vector-database operations to LLM applications via stdio or Server-Sent Events.

162
42
Open Source
GitHub

Installation

1. Prerequisites
• Python 3.9+
• Running Milvus vector database (v2.4 or newer) reachable from the MCP server
• git, pip, and virtualenv (recommended)
2. Clone the repository
git clone https://github.com/zilliztech/mcp-server-milvus.git cd mcp-server-milvus
3. Create & activate a virtual environment (optional but recommended)
python -m venv .venv source .venv/bin/activate # On Windows: .venv\Scripts\activate
4. Install server dependencies
pip install -r requirements.txt
If you plan to run the server in production, also install the extra “prod” dependencies:
pip install -r requirements-prod.txt
5. Configure connection to Milvus
The server looks at the following environment variables:
• MILVUS_HOST – host/IP of the Milvus instance (default: 127.0.0.1)
• MILVUS_PORT – gRPC port of Milvus (default: 19530)
• MILVUS_USER / MILVUS_PASSWORD – (if your Milvus is secured)
• MCP_SERVER_PORT – HTTP port the MCP server should listen on (default: 8080)
Example:
export MILVUS_HOST=milvus.local export MILVUS_PORT=19530 export MCP_SERVER_PORT=8080
6. Start the MCP server
python -m mcp_server_milvus.server
The server will start exposing the MCP REST/GRPC endpoints on the port you configured.
7. Docker alternative (if the project ships a Dockerfile)
docker build -t mcp-server-milvus . docker run -d -p 8080:8080 \ -e MILVUS_HOST=milvus.local \ -e MILVUS_PORT=19530 \ mcp-server-milvus
8. Verify
Open http://localhost:8080/health or call `curl http://localhost:8080/health` to confirm the server is healthy.

Documentation

License: Apache License 2.0
Updated 7/30/2025