DuckDuckGo MCP Server
@shgsousa
A web search tool and API powered by DuckDuckGo, Gradio, and MCP, providing both a user-friendly web interface and Claude Desktop tool integration. It fetches web search results, extracts summaries, and retrieves the full content of web pages in markdown format.
Overview
What is DuckDuckGo MCP Server?
DuckDuckGo MCP Server provides a web-based search interface using the DuckDuckGo search API. It is built with Python and Gradio and is intended for users who want real-time search results with markdown-formatted output.
How to use DuckDuckGo MCP Server?
Run the Docker container (docker run -p 7860:7860 ddg-mcp-server) and access the application at http://localhost:7860. Optionally configure an OpenAI-compatible API key via environment variables (OPENAI_API_URL, ACCESS_TOKEN) to enable AI‑powered content summarization.
Key features of DuckDuckGo MCP Server
- Web‑based search interface using DuckDuckGo
- Real‑time search results with full content
- Markdown‑formatted output
- Configurable number of results
- AI‑powered content summarization (requires API key)
Use cases of DuckDuckGo MCP Server
- Fetching and displaying DuckDuckGo search results in a browser
- Automating search queries with a configurable result count
- Summarizing web content using an OpenAI‑compatible API
FAQ from DuckDuckGo MCP Server
What does the server do?
It provides a Gradio web interface that searches DuckDuckGo and returns results formatted as Markdown. It can optionally summarize content using an external AI API.
How do I run the server?
Build the Docker image (docker build -t ddg-mcp-server .) and run the container (docker run -p 7860:7860 ddg-mcp-server). The application is then available on port 7860.
What dependencies does it require?
The application runs on Python 3.10 and uses Gradio, the DuckDuckGo Search API, BeautifulSoup4, and Markdownify. For summarization, an OpenAI‑compatible API key is needed.
How do I enable AI summarization?
Set the OPENAI_API_URL and ACCESS_TOKEN environment variables (or use a .env file). The default model is gpt-4.1-turbo, configurable in config.py.
Where does the search data come from?
All search results are fetched live from DuckDuckGo's search API. No local search index or cached data is used.
