MCP Server for Deepseek Integration

This repository contains a Model Control Protocol (MCP) server implementation that allows Claude Desktop to use Deepseek models running in Docker.

Prerequisites

• Docker
• Python 3.11 or later
• A Deepseek API key
• Claude Desktop

Installation

1. Clone the repository:

git clone https://github.com/vincentf305/mcp-server-deepseek.git
cd mcp-server-deepseek

2. Install dependencies:

pip install -r requirements.txt

Setup Environment Variables

Create a .env file in the root directory of the project and add the following environment variable:

DEEPSEEK_API_KEY=your_api_key_here

Make sure to replace your_api_key_here with your actual Deepseek API key.

Running the Server

Using Docker

1. Build the Docker image:

docker build -t mcp_server_deepseek .

2. Run the container:

docker run -d \
  --name mcp-server-deepseek \
  -p 8765:8765 \
  -e DEEPSEEK_API_KEY=your_api_key_here \
  mcp-server-deepseek

Running Locally

python -m mcp_server_deepseek.server

Usage with Claude Desktop

1. Ensure you have a Deepseek API key

2. Add the following to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "deepseek-server": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "DEEPSEEK_API_KEY",
        "mcp_server_deepseek"
      ],
      "env": {
        "DEEPSEEK_API_KEY": "your_api_key_here"
      }
    }
  }
}

3. Restart Claude Desktop to load the new configuration

License

MIT License - see the LICENSE file for details