Qwant MCP Server

A Model Context Protocol (MCP) server for the Qwant search engine, inspired by the Kagi MCP implementation.

Features

• Search the web using Qwant's API
• Format results in a structured way for MCP clients
• Simple stdin/stdout communication

Installation

# Clone the repository
git clone https://github.com/yourusername/mcp-qwant.git
cd mcp-qwant

# Build the project
go build -o mcp-qwant ./cmd/mcp-qwant

Usage

# Run the server
./mcp-qwant

The server will start and listen for MCP requests on stdin/stdout.

Configuration

No configuration is required. The server uses Qwant's public API which doesn't require an API key.

Tools

The Qwant MCP server provides four search tools, each specialized for a different type of content:

Web Search Tool

• Name: search_web
• Description: Fetch web results based on one or more queries using the Qwant Search API. Use for general web search and when the user explicitly tells you to 'fetch' or 'search online' results/information.
• Input:

  {
    "queries": ["search query 1", "search query 2"]
  }
  

• Output: Formatted web search results from all queries

News Search Tool

• Name: search_news
• Description: Fetch news results based on one or more queries using the Qwant News Search API. Use when the user is looking for recent news articles or current events.
• Input:

  {
    "queries": ["news query 1", "news query 2"]
  }
  

• Output: Formatted news search results from all queries

Images Search Tool

• Name: search_images
• Description: Fetch image results based on one or more queries using the Qwant Image Search API. Use when the user is looking for visual content, illustrations, or photos.
• Input:

  {
    "queries": ["image query 1", "image query 2"]
  }
  

• Output: Formatted image search results from all queries (includes thumbnail URLs when available)

Videos Search Tool

• Name: search_videos
• Description: Fetch video results based on one or more queries using the Qwant Video Search API. Use when the user is looking for video content, tutorials, or multimedia results.
• Input:

  {
    "queries": ["video query 1", "video query 2"]
  }
  

• Output: Formatted video search results from all queries (includes thumbnail URLs when available)

Development

Running Tests

go test ./...