ScrapeGraphAI/scrapegraph-mcp — reverse-engineered prompt

I want a Python MCP server for ScrapeGraph AI that I can plug into Claude Desktop or Cursor and start using right away.

It should let an assistant scrape a page into useful formats like markdown, html, screenshots, links, images, branding, or summaries. I also want it to extract structured data from a URL based on a plain English prompt, run web searches with a reasonable result limit, start and check multi page crawls, stop or resume them, generate JSON schemas from prompts, and manage scheduled monitors. Please include simple account tools too, like checking credits and viewing request history.

Make it work both locally and with a hosted HTTP connection, using an API key from environment variables or request headers. Keep the setup smooth, add clear docs and examples, and make the error handling solid enough for real use. If there are older tool names that changed, use the current ones and keep the behavior aligned with the latest ScrapeGraph API docs. Look up current MCP docs online if you need to.