mcp-firecrawl

Structured data extraction uses LLM-powered schema-based extraction to pull specific fields from web pages. Define a JSON schema with the data points you need—name, price, status—and the firecrawl_extract tool automatically captures and returns them in structured format from any URL.

Single-page scraping converts web content to markdown automatically. Use firecrawl_scrape with your target URL; it extracts the main content, applies caching for speed, and returns clean markdown output ready for processing or storage.

Single-page scraping extracts content from one URL; site-wide crawling discovers and processes multiple pages automatically. Use firecrawl_scrape for individual pages and firecrawl_crawl for comprehensive site mapping with URL discovery and rate control built in.

Web search with scraping performs searches and optionally extracts content from result URLs in one workflow. The firecrawl_search tool returns results, and you can apply extraction parameters to gather structured data across multiple matching pages without separate requests.

Rate control and caching are built into the firecrawl tools. The skill handles request throttling automatically and caches results to reduce redundant calls, so you don't need custom rate-limiting logic for safe, targeted crawling at scale.

The onlyMainContent parameter focuses extraction on primary article or product text. Enable this flag during scraping or extraction to filter out sidebars, headers, and advertisements, returning only the central content relevant to your task.