Web Search MCP Server

A Model Context Protocol (MCP) server that provides a web search tool using the Tavily API. This server enables AI models to search the internet and retrieve up-to-date information.

Overview

This MCP server implements a single tool:

• search_web_tool: Searches the web using Tavily API and returns relevant search results.

The server acts as a bridge between AI models and the Tavily search engine, allowing models to access current information from the internet during conversations.

Features

• Real-time web search capabilities
• Customizable search parameters:
  • Search topic (general, news, finance)
  • Search depth (basic, advanced)
  • Maximum number of results
  • Time range filtering (day, week, month, year)
  • Domain inclusion/exclusion

Requirements

• Python 3.13+
• uv - Python package installer and resolver
• Tavily API key (sign up at tavily.com)

Installation

1. Clone the repository (if applicable)

2. Set up a virtual environment using uv (optional but recommended):

   uv venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   

3. Install dependencies using uv:

   uv pip sync
   

Configuration

1. Create a .env file by copying the provided template:

   cp env-sample .env
   

2. Add your Tavily API key to the .env file:

   TAVILY_API_KEY=your-api-key-here
   

Usage

Run the server using uv:

uv run web_search_server.py

The server operates using the stdio transport method for MCP communication, making it suitable for integration with various AI systems that support the Model Context Protocol.

Tool Parameters

The search_web_tool accepts the following parameters:

• query (str, required): The search query.
• search_topic (str, optional): The topic of the search. Can be "general", "news", or "finance". Defaults to "general".
• search_depth (str, optional): The depth of the search. Can be "basic" or "advanced". Defaults to "basic".
• max_results (int, optional): The maximum number of results to return. Defaults to 1.
• time_range (str, optional): The time range for the search. Can be "day", "week", "month", or "year". Defaults to None.
• include_domains (list[str], optional): A list of domains to include in the search.
• exclude_domains (list[str], optional): A list of domains to exclude from the search.

Response Format

The tool returns a list of search results, each containing:

• title: The title of the search result
• url: The URL of the search result
• content: The content of the search result
• score: The relevance score of the search result

Error Handling

If an error occurs during the search operation, the tool will return an error message describing the issue.