DDGS MCP Server

A Model Context Protocol (MCP) server that provides DuckDuckGo Search capabilities to AI agents.

Live page: https://ddgs-mcp-server.oriz.in

Features

• search_text: Advanced metasearch using bing, brave, duckduckgo, google, mojeek, yahoo, yandex, wikipedia.

  • Full Content Extraction: Optionally fetch complete page content (not just snippets) for comprehensive context.

• search_news: Find latest updates, releases, and tech news.

Related MCP server: WebRead MCP

Full Content Extraction

For coding agents that need complete context from search results, enable full page content fetching:

Usage

{
  "query": "python async programming tutorial",
  "fetch_full_content": true,
  "max_content_length": 50000,
  "max_results": 5
}

Parameters

Response Structure

When fetch_full_content is enabled, each result includes a full_content field:

[
  {
    "title": "Python Async Programming Guide",
    "href": "https://example.com/python-async",
    "body": "Brief snippet from search results...",
    "full_content": "Complete extracted article text with all paragraphs, code examples, and detailed explanations..."
  }
]

Performance Notes

• Content extraction adds ~1-3 seconds latency per page

• Up to 5 pages are fetched concurrently to minimize total time

• Failed fetches return [Content extraction failed or blocked] without breaking the search

• Uses Trafilatura for high-quality text extraction

Installation & Usage

You can run this server directly using uvx without installing it globally.

VS Code (Claude Desktop / Cline)

Add this to your MCP settings file (e.g., cline_mcp_settings.json or claude_desktop_config.json):

{
  "mcpServers": {
    "ddgs-search": {
      "command": "uvx",
      "args": [
        "ddgs-mcp-server"
      ],
      "disabled": false,
      "alwaysAllow": []
    }
  }
}

Manual Execution

uvx ddgs-mcp-server

Secrets & Configuration

This project technically does not require API keys to run locally, as it scrapes DuckDuckGo. However, for publishing or proxy usage, you should configure your environment.

1. Set up Secrets

Copy the example file:

cp .env.example .env

2. Required Tokens

Development / Publishing

To build and publish this package to PyPI (using the secrets from above):

1. Build:

   pip install build twine
   python -m build

2. Publish (loads secrets from .env if you export them, or prompts you):

   # If using .env variables (PowerShell)
   # $env:TWINE_USERNAME = "__token__"
   # $env:TWINE_PASSWORD = "pypi-..."
   
   python -m twine upload dist/*