WebforAI Text Extractor

You can run a WebforAI MCP Server that extracts plain, markdown-formatted text from any web page URL and serves it to MCP clients. This enables AI models to consume clean web content without HTML noise, while handling errors gracefully and scaling globally.

How to use

Connect to the MCP server from your chosen MCP client to start extracting web content. You provide a target page URL, and the server returns the page text in clean Markdown format, with links and tables converted to readable text and images hidden.

Two common ways to connect are via an HTTP remote MCP endpoint or through a local stdio workflow using a runtime tool like npx. The HTTP option points your client at a deployed server, while the stdio option runs the MCP server locally and talks to it through the command line tool.

Example usage patterns you can follow in your client tooling include feeding the server a page URL to extract content, handling any errors, and displaying the returned Markdown to your end users or AI model prompts.

How to install

Prerequisites: ensure you have Node.js installed and a package manager available. You will also need network access to fetch dependencies during setup.

Step 1: Clone the project repository and navigate into the project folder.

git clone https://github.com/yutakobayashidev/webforai-mcp-server.git
cd webforai-mcp-server

Step 2: Install dependencies using your preferred package manager.

pnpm install

Step 3: Run the development server. This starts locally and serves the MCP endpoint at the default local address.

pnpm dev

Step 4: Open your local endpoint to begin extracting content. The server is available at http://localhost:8787.

Configuration and usage examples

You can connect to the MCP server from the Claude Desktop client by providing a JSON configuration for the MCP server. This lets Claude invoke the extractor as an external tool and pass URLs to it for extraction.

{
  "mcpServers": {
    "webforaiExtractor": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:8787/sse"  // or webforai-mcp-server.your-account.workers.dev/sse
      ]
    }
  }
}

Available tools

extractWebPageText

Accepts a URL to a web page and returns the extracted text content in Markdown format. Converts links to plain text, tables to plain text, and hides images for a clean AI-friendly output.