WebforAI Content Acquisition Module - MCP Endpoint

This is a server implementation based on Cloudflare Workers, adhering to the Model Context Protocol (MCP), engineered to strip down web pages to raw text using the capabilities provided by WebforAI.

🌟 Introduction to WebforAI Capabilities

WebforAI is a sophisticated utility suite designed to render web documents consumable by artificial intelligence constructs. Its functions include:

Rendering HTML into pristine, hierarchical Markdown format
Identifying and isolating key content segments from online documents
Interpreting elements like tabular data, hyperlinks, and embedded visuals intelligently
Preparing sourced web material for streamlined ingestion by AI systems

This particular MCP server harnesses WebforAI's core power to fetch textual data from any specified URL, thereby simplifying the pipeline for injecting web content into AI workflows via the Model Context Protocol.

📋 Core Functionalities

Simplified Interface: Execute content extraction from any web destination via a single programmatic invocation.
Pristine Output: Yields Markdown output that is meticulously formatted, devoid of residual HTML markup.
Resilience: Built-in, robust mechanisms for handling and reporting request failures.
Serverless Infrastructure: Deployed on Cloudflare Workers for global, scale-on-demand operation.
Protocol Compliance: Fully interoperable with various MCP consumers, such as Claude Desktop and Cloudflare's AI Playground.

🚀 Deployment and Setup

Provisioning on Cloudflare Workers

Deploy this service instantly to your Cloudflare Workers environment:

This action will establish your MCP service endpoint, typically accessible at: webforai-mcp-server.<your-account>.workers.dev/sse

Local Development Environment

Obtain a copy of the source code repository: bash git clone https://github.com/yutakobayashidev/webforai-mcp-server.git cd webforai-mcp-server
Install required package dependencies: bash pnpm install
Initiate the local development server: bash pnpm dev
The endpoint will be reachable locally, usually at http://localhost:8787

🔧 Invoking the Text Acquisition Utility

The extractWebPageText function accepts a target URL parameter and returns the retrieved document body formatted as Markdown:

{ "url": "https://example.com/page" }

The resulting payload will contain the extracted text in Markdown format, characterized by: - Hyperlinks rendered as static text. - Tables presented as plain text structures. - Images suppressed from the output.

🔌 Establishing Connectivity with MCP Clients

Integration with Cloudflare AI Playground

Navigate to the Cloudflare AI Playground.
Input the address of your deployed MCP endpoint (webforai-mcp-server.<your-account>.workers.dev/sse).
The text extraction capability will become immediately available for use within the playground interface.

Connecting via Claude Desktop

To integrate this extraction utility with your local Claude Desktop application:

Adhere to the instructions outlined in Anthropic's Quickstart Guide.
Within Claude Desktop's configuration settings, navigate to Settings > Developer > Edit Config.
Update your configuration block with the following structure:

{ "mcpServers": { "webforaiExtractor": { "command": "npx", "args": [ "mcp-remote", "http://localhost:8787/sse" // Substitute with your deployed worker URL if needed ] } } }