Commands

You run an MCP server that combines FastMCP tooling with DeepSeek AI capabilities, enabling you to call MCP tools and interact with AI for conversations, code analysis, and refactoring suggestions.

How to use

After you start the server, you can connect an MCP client to either the local HTTP endpoint or the local stdio endpoint to access built-in tools.

How to install

Prerequisites you need before starting: Node.js installed on your system.

1. Install dependencies and prepare the server.

npm install

2. Start the server.

npm start

Additional setup notes

Set the API key for the DeepSeek AI integration before starting the server. You have a couple of options depending on your OS.

# Windows PowerShell
$env:DEEPSEEK_API_KEY="your-api-key-here"

# Windows CMD
set DEEPSEEK_API_KEY=your-api-key-here

# Linux/Mac
export DEEPSEEK_API_KEY="your-api-key-here"

The server exposes two MCP connection methods you can use from your MCP client.

Configuration and endpoints

HTTP MCP endpoint (recommended): use the local MCP service URL.

Local stdio MCP endpoint: run the included command to start an in-process MCP server.

Troubleshooting

If you encounter issues, perform these checks in order.

• Ensure the DEEPSEEK_API_KEY is correctly set.

• Verify network connectivity if you are using the HTTP endpoint.

• Review console logs for detailed error messages.

• If there is a port conflict, adjust the port in the server configuration.

Notes

The server includes built-in MCP tools for arithmetic and AI-assisted interactions, and a prompt catalog to tailor AI behavior.

Tools and capabilities

• add: compute the sum of two numbers. Parameters: a (number), b (number). Returns a result string.

• deepseek_chat: call DeepSeek AI for intelligent dialogue and code analysis. Parameters include message, system_prompt, model, temperature, and max_tokens. Returns AI reply content.

• get_prompts: retrieve available AI prompt templates by optional template name. Returns template details or the full list.

Environment and language settings

The integration uses Node.js, FastMCP, DeepSeek AI, Zod, Axios, and dotenv to manage configuration and runtime behavior.

Available tools

add

Compute the sum of two numbers with parameters a and b, returning the result as a string.

deepseek_chat

Interact with DeepSeek AI for intelligent dialogue or code analysis. Supports parameters like message, system_prompt, model, temperature, and max_tokens.

get_prompts

Fetch available AI prompt templates, optionally filtering by a specific template name.