Deepseek Thinker

A MCP provider Deepseek reasoning content to MCP-enabled AI Clients, like Claude Desktop. Supports access to Deepseek's CoT from the Deepseek API service or a local Ollama server.

javascript

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ruixingshi-deepseek-thinker-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-thinker-mcp"
      ],
      "env": {
        "API_KEY": "<Your API Key>",
        "BASE_URL": "<Your Base URL>",
        "USE_OLLAMA": "true"
      }
    }
  }
}

You can run the Deepseek Thinker MCP Server to expose Deepseek’s reasoning content to MCP-enabled AI clients. It supports both a local Ollama/OpenAI API workflow and a standalone local server, letting you choose the environment that fits your setup and latency requirements.

How to use

Connect to the Deepseek Thinker MCP Server from your MCP-enabled AI client such as Claude Desktop. You have two practical modes: run via the MCP package with npx for a quick startup, or run a local server with node for a self-contained deployment. You can also enable Ollama mode if you want to route reasoning through a local Ollama instance.

How to install

Prerequisites: ensure you have Node.js and npm installed on your system. Optionally, install Ollama if you plan to use Ollama mode.

# Install dependencies for the project
npm install

# Build the project
npm run build

# Run the service (standalone local server)
node build/index.js

Configuration and usage examples

Configure your MCP client to connect to the Deepseek Thinker MCP Server using the following examples. You will typically place these in your client’s MCP configuration file.

{
  "mcpServers": {
    "deepseek_thinker": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-thinker-mcp"
      ],
      "env": {
        "API_KEY": "<Your API Key>",
        "BASE_URL": "<Your Base URL>"
      }
    }
  }
}

Using Ollama mode

If you want to route reasoning through Ollama, enable Ollama mode in your MCP config. The same npx command starts the MCP server but sets an environment flag to use Ollama.

{
  "mcpServers": {
    "deepseek_th thinker": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-thinker-mcp"
      ],
      "env": {
        "USE_OLLAMA": "true"
      }
    }
  }
}

Running a local server

For a self-contained local server, run the MCP server directly via Node.js and point it at your built artifact. This approach does not rely on an external npx invocation.

{
  "mcpServers": {
    "deepseek_thinker_local": {
      "command": "node",
      "args": [
        "/your-path/deepseek-thinker-mcp/build/index.js"
      ],
      "env": {
        "API_KEY": "<Your API Key>",
        "BASE_URL": "<Your Base URL>"
      }
    }
  }
}

Security and secrets

Keep your API keys and base URLs secure. Do not commit credentials to version control. Use environment variable management appropriate for your deployment target.

Notes and troubleshooting

If you encounter a timeout like MCP error -32001, this usually means the Deepseek API response is slow or the output is too large for the server to handle within the allowed window. Adjust prompt length or retry with a shorter reasoning payload.

Available tools

get-deepseek-thinker

Perform reasoning using the Deepseek model and receive a structured reasoning process as the output

Built by

VeilStrat

AI signals for GTM teams