DuckDuckGo

Provides fast, free web search from DuckDuckGo via MCP with search, crawl, and research capabilities.

javascript

1

GitHub Stars

javascript

Language

4 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": {
    "nipurn123-duckduckgo-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "duckduckgo-mcp"
      ]
    }
  }
}

You can run a DuckDuckGo-based MCP server that provides fast, free web search capabilities through MCP tools. It enables you to perform quick searches, crawl result pages in parallel, and rank content by relevance without requiring API keys or rate limits.

How to use

Use this MCP server with an MCP client to access three core tools: search, search_and_crawl, and research. You can run the server locally and connect to it through your MCP client to perform fast searches, fetch content from multiple sources in parallel, and obtain ranked results that help you answer questions efficiently.

How to install

Prerequisites: you need Node.js and npm installed on your system.

# From source: clone the project
git clone https://github.com/yourusername/duckduckgo-mcp.git
cd duckduckgo-mcp

# Install dependencies
npm install

# Build the project
npm run build

Next, configure your MCP client to run the MCP server. Here are two common runtime options shown in the available configurations.

## Additional configuration notes

You can start the MCP server using a JavaScript runtime pointing to the built entry file. For example, after building, you can run the server with Node by specifying the path to the built index.

{ "mcpServers": { "duckduckgo": { "command": "node", "args": ["/path/to/duckduckgo-mcp/build/index.js"] } } }


Alternatively, you can use a convenience runner with npx if you prefer not to install locally. For example, you can configure the MCP client to launch with npx and the package name directly.

Tools

This MCP server provides three tools to interact with web content simply and efficiently.

1. search

Quick web search. Returns titles, URLs, and descriptions for your query.

2. search_and_crawl

Search and crawl: fetch multiple result pages in parallel and retrieve full content from each source.

3. research ⭐

Best for answering questions. Searches, crawls in parallel, then ranks results by relevance to your question using a defined scoring system.

How it works

The server leverages DuckDuckGo HTML endpoints to fetch static HTML, avoids CAPTCHA issues common with other search engines, crawls results in parallel for speed, and extracts the main content while removing navigation and ads to improve extraction quality.

Source Authority Scores

High-authority domains receive higher rankings based on predefined domain scores to influence result quality.

Development notes

You can build and test the MCP server locally. Use the included commands to install dependencies, build, and then test the server through its entry point.

Security and limitations

Be aware that some sites block scrapers (403 errors) and content quality can vary by source. This approach may not be ideal for JavaScript-heavy single-page applications.

Available tools

search

Performs a quick web search and returns titles, URLs, and descriptions for the given query.

search_and_crawl

Executes a search and crawls the resulting pages in parallel to obtain full content from each source.

research

Conducts a search and parallel crawl, then ranks results by relevance to the provided question using a scoring model.