Multi Fetch MCP Server

Example Use Cases

• 1

  Fetch and extract article content and metadata from web pages (with Readability-based extraction) for downstream reasoning or summarization.

• 2

  Retrieve very large web documents incrementally or in parts, using byte-level chunking and startCursor controls to avoid model context limits.

• 3

  Reliable access to JavaScript-rendered or anti-scraping pages by falling back to a headless browser mode with proxy and browser automation options.

Description

This project implements a Model Context Protocol (MCP) server and client that provide web-fetching tools for AI agents via the stdio transport. It supports multiple fetch modes (node fetch and headless browser), intelligent content extraction (Readability), metadata extraction, proxy support, and byte-level content chunking to handle large pages. Built-in debug logging, bilingual i18n (en/zh), and tools integration (e.g., Claude desktop) make it easy to integrate into agent workflows that need reliable web retrieval and progressive content delivery.