Web Fetch

Web Fetch is an MCP server that enables AI assistants to retrieve, render, and convert online content across the web. Developed in TypeScript using Playwright, this server connects directly to external websites, web pages, and raw text-based endpoints to extract web data for developers, researchers, and automated LLM workflows. Instead of relying solely on simple HTTP requests, Web Fetch provides headless browser rendering to execute JavaScript on modern Single Page Applications and complex web interfaces, returning complete HTML structures. For content processing and documentation tasks, the server transforms web pages into structured Markdown, retaining elements like tables and definition lists while also providing options to automatically strip away navigation menus, headers, and footers for cleaner article summaries. It also allows direct access to raw text formats such as JSON, XML, CSV, and TSV without browser overhead. By exposing specialized tools for text retrieval, headless HTML rendering, and Markdown extraction, Web Fetch allows AI models to inspect real-time web documents, archive web resources, and conduct structured data extraction directly inside supported client applications.

Category: Browser & Web Automation

Tags: data-extraction, html, web crawling, web scraping

Visit Web Fetch

How to install and configure Web Fetch

1. Install the package globally via npm or run it directly using npx: bash npm install -g mcp-server-fetch-typescript 2. Open your Claude Desktop configuration file: - macOS: ~/Library/Application Support/Claude/claude_desktop_config.json - Windows: %APPDATA%/Claude/claude_desktop_config.json 3. Add mcp-server-fetch-typescript to the mcpServers configuration: json { "mcpServers": { "mcp-server-fetch-typescript": { "command": "npx", "args": [ "-y", "mcp-server-fetch-typescript" ] } } } 4. Alternatively, clone the repository from GitHub, run npm install and npm run build, then configure the server command to execute node targeting /path/to/mcp-server-fetch-typescript/build/index.js. 5. Save the configuration and restart Claude Desktop.

What you can do with Web Fetch

• Fetching and executing client-side JavaScript on single-page applications using Playwright to retrieve fully rendered HTML for data extraction. - Converting web articles and documentation into clean Markdown format while preserving structural elements like tables and definition lists. - Extracting main article body text while stripping away headers, footers, and navigation bars using the markdown summary tool. - Retrieving raw structured data from online endpoints including raw text, JSON, XML, CSV, or TSV without browser rendering overhead.

Key facts

• https://github.com/tatn/mcp-server-fetch-typescript
• Browser & Web Automation, Web Search & Research
• data-extraction, html, web crawling, web scraping

Part of MCP Servers

What can Web Fetch do?

Web Fetch allows AI assistants to fetch and transform web content. It includes tools to retrieve raw text from data files like JSON or CSV, execute JavaScript with Playwright to get rendered HTML from single-page applications, convert full pages into formatted Markdown, and extract clean summaries that omit navigation bars and footers.

Which MCP clients work with Web Fetch?

Web Fetch works with any client compatible with the Model Context Protocol, including Claude Desktop. You can configure it directly inside client configuration files using npx or by running the built TypeScript project with Node.js. It can also be inspected using the official MCP inspector tool.

How do I install Web Fetch?

You can install Web Fetch globally with npm install -g mcp-server-fetch-typescript or run it directly through npx. To connect it to Claude Desktop, add an entry under mcpServers in your claude_desktop_config.json file pointing to the npx command with the -y and mcp-server-fetch-typescript arguments, then restart the application.

How does Web Fetch render JavaScript on dynamic websites?

Web Fetch utilizes Playwright for headless browser rendering via the get_rendered_html tool. When a target URL points to a modern web application or single-page app that requires client-side JavaScript execution, Playwright renders the page in a headless environment and returns the fully evaluated DOM HTML to the client.