MCP Web Search Tool

MCP Web Search Tool is an MCP server that provides AI assistants with live web search, full-page reading, and source citation capabilities. It connects AI models to search engines including the Brave Search API and DuckDuckGo, operating over standard I/O (stdio) transport. Developers, researchers, and assistant users implement this server to allow LLMs to retrieve real-time facts, current news, and web images without relying on external web-scraping frameworks. The server provides dedicated tools for conducting structured searches across the web, fetching recent news articles, retrieving image results, and safely reading webpage content by result ID or direct URL. Its built-in webpage fetcher extracts clean readable text, strips boilerplate scripts and styles, and guards against server-side request forgery by blocking private IP ranges. Built with Node.js, the server uses pluggable search providers and LRU caching with TTL expiration to optimize performance. When an optional Brave Search API key is supplied, users access web, news, and image tools; otherwise, DuckDuckGo functions as a keyless fallback for standard text searches.

Category: Web Search & Research

Tags: brave, internet, search, web

Visit MCP Web Search Tool

How to install and configure MCP Web Search Tool

1. Clone the repository and install project dependencies: bash git clone https://github.com/gabrimatic/mcp-web-search-tool.git cd mcp-web-search-tool npm install 2. Create an environment configuration file: bash cp .env.example .env Edit .env to supply BRAVE_API_KEY if you have one. 3. Build and run the server: bash npm run build npm start 4. Or build and run using Docker: bash docker build -t mcp-web-search . docker run --rm -i -e BRAVE_API_KEY mcp-web-search 5. Register the command or Docker container in your MCP client configuration file for Claude Desktop, Cursor, or Windsurf.

What you can do with MCP Web Search Tool

• Querying real-time sports results, weather updates, or current events through live web search with freshness filters and country-specific constraints. - Retrieving recent news articles with publication timestamps and publisher metadata using the Brave Search provider. - Searching for relevant online imagery and obtaining thumbnail references directly inside conversational AI workflows. - Extracting readable article content and outbound links from search result IDs while automatically stripping script tags and navigation boilerplate. - Running keyless local web searches through the built-in DuckDuckGo provider fallback when an API key is unavailable.

Key facts

• https://github.com/gabrimatic/mcp-web-search-tool
• Web Search & Research
• brave, internet, search, web

Part of MCP Servers

How do I install MCP Web Search Tool?

You can install the server by cloning the GitHub repository, installing dependencies with npm install, creating a .env file with optional configuration, and compiling the project using npm run build. It can then be started via npm start or run in a Docker container using stdio transport.

What can MCP Web Search Tool do?

The server offers five tools: web_search, news_search, image_search, fetch_url, and list_providers. It retrieves ranked web results, news articles with publication details, image thumbnails, and page reader output that extracts readable text and links while stripping non-content HTML elements.

Is an API key required to use MCP Web Search Tool?

A Brave Search API key is optional for general web search, as DuckDuckGo operates as an automatic fallback without requiring a key. However, accessing the news_search and image_search tools requires a valid Brave Search API key configured in the environment.

Which MCP clients work with MCP Web Search Tool?

Any client supporting standard Model Context Protocol stdio transport can connect to this server. Documented client environments include Claude Desktop, Claude Code, Codex, Visual Studio Code, Cursor, and Windsurf, which execute the server locally or through Docker containers.

Is MCP Web Search Tool open source?

Yes, MCP Web Search Tool is open-source software released under the permissive MIT License. Users are permitted to review the TypeScript source code, modify functionality, and integrate it into custom environments according to the license terms.