Firecrawl MCP

What it does

Firecrawl MCP connects Claude to the Firecrawl API — a purpose-built web crawling service that handles JavaScript rendering, anti-bot bypasses, rate limiting, and structured extraction at scale. Unlike simple HTTP fetchers, Firecrawl can traverse an entire site, following links and returning cleaned content from every page.

This makes it the go-to tool for competitive research, content audits, data extraction projects, and any task requiring data from more than a handful of pages. The free tier handles light usage; API keys are needed for large crawls.

• Scrape individual pages with JavaScript rendering
• Crawl entire websites following internal links
• Map site URL structure without fetching full content
• Extract structured data using LLM-powered schemas
• Search the web and return scraped results

How to connect

Set your Firecrawl API key as an environment variable, then connect:

export FIRECRAWL_API_KEY="fc-your-key-here"
claude mcp add firecrawl -- npx -y firecrawl-mcp

Available tools

• scrape_url — scrape a single URL with JavaScript rendering
• crawl_website — crawl a site following all internal links
• map_website — return the URL map of a site without full content
• search_web — search and return scraped results
• extract_structured_data — extract data using a JSON schema prompt

Example usage

Ask Claude to extract product data from a competitor's catalogue or map a documentation site:

# User prompt
Crawl competitor.com/pricing — visit every plan page
and extract the plan name, monthly price, and feature
list as structured JSON.

map_website("https://competitor.com/pricing")
→ returns 12 pricing page URLs
extract_structured_data({
  url: "https://competitor.com/pricing/pro",
  schema: { plan: "string", price: "number", features: "array" }
})
→ returns clean JSON for each page