Firecrawl CLI & Skill

<p>Firecrawl CLI & Skill gives AI agents five web capabilities in a single install: scrape, search, browse, crawl, and map. Run <code>npx skills add firecrawl/cli</code> and your Claude Code, Codex, or Gemini CLI agent can pull clean markdown from any page, search the web and scrape results simultaneously, launch cloud browser sessions for interactive sites, recursively crawl entire domains, and discover every URL on a site.</p> <p>The output is optimized for LLM consumption — clean markdown, not raw HTML. Firecrawl handles the ugly parts: JavaScript rendering, anti-bot bypasses, dynamic content loading, and cookie walls. Your agent gets structured text it can actually reason over.</p> <p>The skill format is what sets this apart from using Firecrawl as a standalone API. When installed as a skill, the agent <em>learns</em> when and how to use each capability autonomously. A research agent will automatically search and scrape when building a report. A data extraction agent will use the browser capability when it encounters a login wall. The agent decides — you don't have to prompt for it.</p> <p>Firecrawl is also available as an MCP server (<code>claude mcp add firecrawl -- npx -y firecrawl-mcp</code>) for agents that support the Model Context Protocol. The main Firecrawl repo has 90,000+ GitHub stars, making it one of the most popular web scraping tools in the AI ecosystem.</p> <p>Pricing starts with a free tier (500 credits/month), with paid plans from $19/month (3,000 credits) up to $399/month (500,000 credits). One scrape costs 1 credit. One crawled page costs 1 credit.</p> <p>Compatible agents: Claude Code (official first-party plugin), Codex CLI, Gemini CLI, Antigravity, OpenCode, and Cursor (via MCP).</p>