Open Deep Research MCP Server

This is a Model Context Protocol (MCP) server that wraps the HuggingFace Open Deep Research project, built on their smolagents library, making the project available for use with MCP clients, such as AI assistants.

This is an early work-in-progress, has limited testing, and requires access to API's that will be replaced in the near future. For now, the following API keys are required to be set in the environment or in a .env file (see .env.example):

OpenAI API Key: https://platform.openai.com/
HuggingFace API Key: https://huggingface.co/
SERP API Key: https://serpapi.com/

Setup and Installation

Simply run:

make

To create the virtual environment and install dependencies.

Running the Server

Use the VSCode launch configuration, or run manually:

Defaults to stdio transport:

uv run -m mcp_server.start

For SSE transport:

uv run -m mcp_server.start --transport sse --port 6020

The SSE URL is:

http://127.0.0.1:6020/sse

Client Configuration

To use this MCP server in your setup, consider the following configuration:

Stdio

{
  "mcpServers": {
    "open-deep-research": {
      "command": "uv",
      "args": ["run", "-m", "mcp_server.start"]
    }
  }
}

SSE

{
  "mcpServers": {
    "open-deep-research": {
      "command": "http://127.0.0.1:6020/sse",
      "args": []
    }
  }
}

Open Deep Research MCP Server