Deepseek

🚀 Deep Dive into MCP Server

A Model Context Protocol (MCP) server for DeepSeek API, enabling seamless integration of DeepSeek's powerful language models with MCP-compatible applications (such as Claude Desktop).

Anonymously Use DeepSeek API - Only See the Proxy on the Other Side

🚀 Quick Start

This MCP server allows you to integrate DeepSeek's language models with MCP-compatible applications easily.

📦 Installation

Automatic Installation via Smithery

To automatically install the DeepSeek MCP server for Claude Desktop via Smithery:

npx -y @smithery/cli install @dmontgomery40/deepseek-mcp-server --client claude

Manual Installation

npm install -g deepseek-mcp-server

Usage with Claude Desktop

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "deepseek": {
      "command": "npx",
      "args": ["@dmontgomery40/deepseek-mcp-server"],
      "env": {}
    }
  }
}

✨ Features

• DeepSeek R1 Model: By default, it uses DeepSeek's R1 model (deepseek-reasoner), offering state-of-the-art inference and performance for general tasks.
• Context Management: Handles all context management and message formatting, allowing you to focus on actual interactions rather than technical details.
• Support for Complex Conversations: Suitable for scenarios such as multi-step reasoning, troubleshooting, and detailed technical discussions.

💻 Usage Examples

Local Testing with MCP Inspector

You can use the MCP Inspector tool to test the server locally:

Basic Usage

1. Build the server:

npm run build

2. Run the server using MCP Inspector:

# Make sure to specify the full path to the built file
npx @modelcontextprotocol/inspector node ./build/index.js

The Inspector will open in the browser and connect to the server via stdio transmission. You can:

• View available tools.
• Test chat completions with different parameters.
• Debug server responses.
• Monitor server performance.

⚠️ Important Note

By default, the server uses DeepSeek's R1 model (deepseek-reasoner), offering state-of-the-art inference and performance for general tasks.

📄 License

This project is licensed under the MIT license.