Firecrawl

This gives Claude real web scraping and browser automation through the Firecrawl CLI. It's the escalation path you want: search when you don't have a URL yet, scrape when you do, map to find specific subpages on large sites, crawl for bulk extraction, and browser when you hit interaction walls like logins or pagination. The outputs are markdown optimized for LLM context windows, which matters when you're feeding scraped content back into conversations. It writes everything to .firecrawl/ by default and includes parallelization support up to your account's concurrency limit. The workflow guidance is solid, especially the part about not re-scraping URLs you already fetched with search.

Install to Claude Code

npx -y skills add pedronauck/skills --skill firecrawl --agent claude-code