MCP Web Search Server

A Model Context Protocol (MCP) server that provides multi-engine web search capabilities with intelligent content extraction using a hybrid approach.

Features

• 🔍 Hybrid Search Engine: Fast goquery-based search results + intelligent chromedp content extraction
• 🌐 Multi-Engine Support: Bing, Brave, and DuckDuckGo with smart fallback mechanisms
• 📄 Intelligent Content Extraction: Advanced article parsing with multiple content selectors
• 🚀 Concurrent Processing: Parallel content extraction with rate limiting
• 🤖 AI-Ready Summaries: Aggregated content optimized for AI analysis and summarization
• 🛠️ MCP Protocol: Full compliance with Model Context Protocol specification

Installation

Via go install

go install github.com/liliang-cn/mcp-websearch-server@latest

From Source

git clone https://github.com/liliang-cn/mcp-websearch-server
cd mcp-websearch-server
go build -o mcp-websearch-server

Usage

Standalone

# Show help
mcp-websearch-server --help

# Run the server (stdio mode)
mcp-websearch-server

Integration with Claude Desktop

Add to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "websearch": {
      "command": "mcp-websearch-server"
    }
  }
}

If installed via go install, make sure ~/go/bin is in your PATH.

Using with Claude Code (Skills)

This project includes a SKILL.md definition that can be used with Claude Code. This skill provides specialized instructions for Claude to effectively use the web search tools for deep research and fact-checking.

To use it:

1. Ensure the MCP server is configured in your claude_desktop_config.json (as above).
2. Copy SKILL.md to your skills directory:

   mkdir -p ~/.claude/skills
   cp SKILL.md ~/.claude/skills/web-search.md
   

3. You can now use the skill in Claude Code via /web-search or by asking natural language questions that trigger the skill.

Available Tools

🔍 websearch_basic

Basic web search returning titles, URLs and snippets from a single search engine using the hybrid approach.

Parameters:

• query (string, required): The search query
• max_results (int, optional): Maximum results to return (default: 10)

📄 websearch_with_content

Web search with intelligent content extraction from result pages using chromedp.

Parameters:

• query (string, required): The search query
• max_results (int, optional): Maximum results to return (default: 5)
• extract_content (bool, optional): Extract full page content (default: true)

🚀 websearch_multi_engine

Comprehensive search across multiple engines (Bing, Brave, DuckDuckGo) with content extraction.

Parameters:

• query (string, required): The search query
• max_results (int, optional): Maximum results to return (default: 3)
• engines (array, optional): Search engines to use ["bing", "brave", "duckduckgo"] (default: all)

🤖 websearch_ai_summary

Search and return AI-ready aggregated content optimized for analysis and summarization.

Parameters:

• query (string, required): The search query
• max_results (int, optional): Maximum results to return (default: 3)

Returns: Formatted markdown content with proper structure for AI processing.

Architecture

mcp-websearch-server/
├── main.go                     # Entry point with CLI flags
├── mcp/                        # MCP protocol implementation
│   └── server.go              # MCP server and tool registration
├── search/                     # Search engine implementations
│   ├── interface.go           # Common interfaces
│   ├── hybrid_searcher.go     # Hybrid multi-engine searcher
│   ├── multi_engine.go        # Basic multi-engine orchestration
│   ├── bing_goquery.go        

... [View full README on GitHub](https://github.com/liliang-cn/mcp-websearch-server#readme)