WebforAI

You can run a WebforAI MCP server that extracts plain text from any web page URL and returns clean, Markdown-formatted output optimized for AI models. This server is designed to be MCP-compatible, so you can connect using common MCP clients to feed web content into your AI workflows with robust error handling and simple usage.

How to use

Connect to your deployed MCP server from your chosen MCP client and start extracting text from web pages. Use the extraction tool to request the content of a URL, and receive a Markdown-formatted representation with links converted to plain text, tables rendered as plain text, and images omitted to keep the result clean for AI consumption. You can run the server locally for development or deploy it to Cloudflare Workers for global availability.

How to install

Prerequisites you need before starting:

• Node.js and npm or pnpm installed on your machine
• Access to a cloud hosting service if you plan to deploy (optional)

Local development steps you can follow to run the server locally:

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

Your local server will be available at http://localhost:8787

## Additional content

Configuration and usage notes: the MCP client connects using a stdio configuration that runs the MCP server command directly on your machine. The example provided shows how to invoke the MCP remote tool using npx to point at a local or remote server. You can also connect using an HTTP-based MCP endpoint if you deploy the server to a remote URL.

Example HTTP/stdio endpoints shown in the source illustrate both a local development URL and a remote deployment URL you can target in your MCP client configuration.

Troubleshooting tips: verify that the server is reachable at the configured URL, ensure the MCP client points to the correct /sse path, and check error messages from the client to identify issues with URL access or command execution.

## Available tools

### extractWebPageText

Tool that accepts a URL to a web page and returns the extracted text content in Markdown format with links as plain text, tables as plain text, and images hidden.