Firecrawl Browser

When a site hides content behind clicks, forms, or login walls and regular scraping fails, this gives Claude a persistent browser session in the cloud. It's part of Firecrawl's escalation pattern: try search first, then scrape, then map or crawl, and finally reach for interact as a last resort. You get commands like click, fill, type, and scroll, plus snapshot with element refs so Claude can target the right buttons. Sessions persist across reconnects, which means you can log in once and keep working. The docs are explicit about never using this for web searches, which is the right call since spinning up a browser when a search API works is wasteful.

Install to Claude Code

npx -y skills add firecrawl/cli --skill firecrawl-browser --agent claude-code