Wraps Firecrawl's v2 API to scrape, crawl, and extract structured data from websites without getting blocked. Seven endpoints cover single pages, full site crawls, batch jobs, and autonomous agents that can navigate multi-step workflows. Handles the annoying stuff: JavaScript rendering, CAPTCHA bypass, PDF parsing, and converting everything to markdown or JSON that LLMs can actually use. The skill documentation warns about 10 specific errors including v2.0.0 breaking changes and race conditions, which suggests the author has burned their hands already. Stealth mode costs 5 credits per request, but auto mode only charges if basic scraping fails first. Useful when you need reliable web data for RAG pipelines or agent tools, not just one-off scrapes you could curl.
npx skills add https://github.com/jezweb/claude-skills --skill firecrawl-scraper