URL Content Extractor

A local MCP server providing the fetch_url_content tool. It fetches HTTP/HTTPS pages, removes common non-content HTML elements, and returns text with a source URL.

Requirements

Python 3.14 or later. This release uses the official MCP Python SDK 2.x.

Run

uvx url-content-extractor

The server uses stdio and waits for an MCP client; it is not an interactive terminal application.

MCP client configuration

{
  "mcpServers": {
    "url-content-extractor": {
      "command": "uvx",
      "args": ["url-content-extractor"]
    }
  }
}

Tool

fetch_url_content(url: str) accepts an HTTP or HTTPS URL. Requests follow redirects and use a 30-second timeout for network operations. Normal results are limited to 5000 characters including the source URL. Failures return a short Chinese error message.

This is basic HTML text extraction, not a browser: JavaScript is not executed, and navigation or advertising text may remain. The return limit does not limit download size. URLs are fetched from the machine running the server; use it with trusted callers and URLs.

0.1.4

Updated implementation based on MCP Python SDK 2.x; URL validation, content-type checks, text cleanup, concise errors, and a packaged stdio command entry point.