Firecrawl gives Claude command-line access to web scraping, crawling, and structured data extraction through the Printing Press CLI. You get batch scraping across multiple URLs, deep research operations that run multi-step queries, and LLM-powered extraction that turns messy HTML into clean JSON schemas. The agent mode flags (--agent, --select, --deliver) are built specifically for programmatic use, so Claude can pipe results, filter huge responses down to essential fields, and route output to files or webhooks without manual plumbing. Auth is token-based, stored locally or via environment variable. Useful when you need to pull live web data into an agent workflow rather than hardcoding scraped snapshots.
npx skills add https://github.com/mvanhorn/printing-press-library --skill pp-firecrawl