概要
What is mult-fetch-mcp-server?
mult-fetch-mcp-server is an MCP (Model Context Protocol)-compliant server that enables AI assistants to fetch and process web content. It supports multiple scraping methods (HTML, JSON, Markdown, plain text), intelligent content extraction, automatic chunking of large content, and bilingual English/Chinese interface.
How to use mult-fetch-mcp-server?
Install globally via pnpm add -g @lmcc-dev/mult-fetch-mcp-server or run directly with npx @lmcc-dev/mult-fetch-mcp-server. Integrate with Claude Desktop by adding a server configuration in claude_desktop_config.json using the npx method. Then restart Claude and use tools like fetch_html, fetch_json, fetch_txt, fetch_markdown, or fetch_plaintext.
Key features of mult-fetch-mcp-server
- Multiple web scraping methods: HTML, JSON, text, Markdown, plain text
- Intelligent content extraction using Mozilla's Readability library
- Automatic content size management and chunked retrieval
- Bilingual internationalization (English and Chinese)
- Support for Standard Input/Output (Stdio) transport
- Browser automation enhancements: scrolling, cookie management, selector waiting
Use cases of mult-fetch-mcp-server
- AI assistants fetching web page text for summarization or analysis
- Retrieving JSON data from APIs within conversational workflows
- Extracting clean Markdown from articles for further processing
- Accessing large content that exceeds context limits via chunked retrieval
FAQ from mult-fetch-mcp-server
What transport method does mult-fetch-mcp-server use?
It implements the Standard Input/Output (Stdio) transport method as defined by the MCP specification.
What are the runtime dependencies?
It requires Node.js (version indicated by Node.js badge) and is available as an npm package. For browser-based fetching, it manages a browser instance internally.
How can I set the language?
Set the MCP_LANG environment variable to en (English) or zh (Chinese). Default falls back to OS language then English.
How does content chunking work?
Large content is automatically split into manageable chunks. Clients can request specific chunks using parameters like startCursor and contentSizeLimit, and the demo client supports --all-chunks for sequential retrieval.
How do I enable debug logs?
Pass "debug": true in the tool call parameters. Debug messages are written to stderr and also to a log file when debug mode is active.
