enactfirecrawl
Scrape, crawl, search, and extract structured data from websites using Firecrawl API - converts web pages to LLM-ready markdown
When & Why to Use This Skill
This Claude skill leverages the Firecrawl API to seamlessly scrape, crawl, and search the web, converting complex websites into clean, LLM-ready markdown. It empowers AI agents to bypass anti-bot protections and extract structured data with high precision, making it an essential tool for automated web data processing and real-time knowledge retrieval.
Use Cases
- Market Intelligence: Automatically crawl competitor websites to track product updates, pricing changes, and industry news for strategic analysis.
- AI Knowledge Base Construction: Scrape technical documentation or blogs and convert them into LLM-friendly markdown to power RAG (Retrieval-Augmented Generation) systems.
- Structured Data Extraction: Utilize AI-driven extraction to pull specific data points like job listings, e-commerce product details, or real estate info directly into structured JSON format.
- Real-time Web Research: Perform deep web searches and retrieve full-page content to provide up-to-date answers and comprehensive reports on trending topics.
| name | enact/firecrawl |
|---|---|
| version | 1.2.1 |
| description | Extract structured data from a page using AI |
| enact | "2.0" |
| from | python:3.12-slim |
| secret | true |
| command | python /workspace/firecrawl.py ${action} ${url} ${formats} ${limit} ${only_main_content} ${prompt} ${schema} |
| timeout | 300s |
| license | MIT |
| readOnlyHint | true |
| openWorldHint | true |
| type | string |
| - scrape | Extract content from a single URL |
| - crawl | Discover and scrape all subpages of a website |
| - map | Get all URLs from a website (fast discovery) |
| - search | Search the web and get scraped results |
| - extract | Extract structured data using AI |
| default | "" |
| - For map | Search query to filter URLs |
| - For extract | Natural language instruction for what to extract |
| url | "https://news.ycombinator.com" |
| action | "extract" |
| limit | 5 |
| prompt | "Extract the top 5 news headlines with their URLs and point counts" |
Firecrawl Web Scraping Tool
A powerful web scraping tool that uses the Firecrawl API to convert websites into clean, LLM-ready markdown and extract structured data.
Features
- Scrape: Extract content from a single URL as markdown, HTML, or with screenshots
- Crawl: Automatically discover and scrape all accessible subpages of a website
- Map: Get a list of all URLs from a website without scraping content (extremely fast)
- Search: Search the web and get full scraped content from results
- Extract: Use AI to extract structured data from pages with natural language prompts
Setup
- Get an API key from firecrawl.dev
- Set your API key as a secret:
enact env set FIRECRAWL_API_KEY <your-api-key> --secret --namespace enact
This stores your API key securely in your OS keyring (macOS Keychain, Windows Credential Manager, or Linux Secret Service).
Usage Examples
CLI
Scrape a single page
enact run enact/firecrawl -a '{"url": "https://example.com", "action": "scrape"}'
Crawl an entire documentation site
enact run enact/firecrawl -a '{"url": "https://docs.example.com", "action": "crawl", "limit": 20}'
Map all URLs on a website
enact run enact/firecrawl -a '{"url": "https://example.com", "action": "map"}'
Search the web
enact run enact/firecrawl -a '{"url": "latest AI developments 2024", "action": "search", "limit": 5}'
Extract structured data with AI
enact run enact/firecrawl -a '{"url": "https://news.ycombinator.com", "action": "extract", "prompt": "Extract the top 10 news headlines with their URLs"}'
Extract with a JSON schema
enact run enact/firecrawl -a '{
"url": "https://example.com/pricing",
"action": "extract",
"prompt": "Extract pricing information",
"schema": "{\"type\":\"object\",\"properties\":{\"plans\":{\"type\":\"array\",\"items\":{\"type\":\"object\",\"properties\":{\"name\":{\"type\":\"string\"},\"price\":{\"type\":\"string\"}}}}}}"
}'
MCP (for LLMs/Agents)
When using this tool via MCP, call enact__firecrawl with these parameters:
Scrape a single page
Call with url set to the target URL and action set to "scrape".
Crawl a documentation site
Call with url, action set to "crawl", and limit to control the maximum number of pages.
Map all URLs on a website
Call with url and action set to "map" to discover all URLs without scraping content.
Search the web
Call with url set to your search query (e.g., "latest AI news") and action set to "search". Use limit to control result count.
Extract structured data with AI
Call with url, action set to "extract", and prompt describing what data to extract. Optionally provide a schema for structured output.
Output
The tool returns JSON with:
- markdown: Clean, LLM-ready content
- metadata: Title, description, language, source URL
- extract: Structured data (for extract action)
- links: Discovered URLs (for map action)
API Features
Firecrawl handles the hard parts of web scraping:
- Anti-bot mechanisms
- Dynamic JavaScript content
- Proxies and rate limiting
- PDF and document parsing
- Screenshot capture