Firecrawl Connector

The Firecrawl connector integrates Agentcy with the Firecrawl web scraping and extraction API. It provides 2 live tools for scraping web pages and extracting structured data using LLM-powered extraction.

Authentication

The connector authenticates using a Firecrawl API key from firecrawl.dev.

{
  "name": "Firecrawl",
  "source_type": "firecrawl",
  "config": {
    "api_key": "fc-..."
  }
}

Variable	Required	Description
api_key	Yes	Firecrawl API key

Live Tools (2)

Tool	Description	Arguments
firecrawl_scrape_url	Scrape a web page and return clean markdown content	url, formats?, wait_for?, timeout?
firecrawl_extract	Extract structured data from a URL using LLM-powered extraction	url, schema, prompt?

Example: Agent Using Firecrawl Tools

User: "Scrape the pricing page of competitor.com"

Agent: I'll scrape the page content.
→ Calls firecrawl_scrape_url(url: "https://competitor.com/pricing")
→ Returns clean markdown content of the pricing page
→ Summarizes: "Competitor offers 3 tiers:
  - Starter: $29/mo (5 users, 10GB)
  - Pro: $99/mo (25 users, 100GB)
  - Enterprise: Custom pricing"

User: "Extract all product names and prices from this catalog page"

Agent: I'll extract structured data.
→ Calls firecrawl_extract(
    url: "https://store.example.com/catalog",
    schema: {"products": [{"name": "string", "price": "number", "in_stock": "boolean"}]},
    prompt: "Extract all product listings with name, price, and availability"
  )
→ Returns structured JSON with extracted products

ETL Ingestion

This connector does not support ETL ingestion. All tools operate as live API calls to the Firecrawl service.

Configuration Reference

{
  "name": "Firecrawl",
  "source_type": "firecrawl",
  "config": {
    "api_key": "fc-..."
  }
}

Field	Type	Default	Description
api_key	string	required	Firecrawl API key

Troubleshooting

Error	Cause	Fix
401 Unauthorized	Invalid API key	Verify your Firecrawl API key
402 Payment Required	Credits exhausted	Top up credits on firecrawl.dev
Scrape failed	Target page blocks scraping or uses heavy JavaScript	Try adding wait_for parameter to wait for dynamic content
Timeout	Page took too long to load	Increase the timeout parameter
Extraction failed	LLM could not match the schema to page content	Simplify the schema or add a more specific prompt
URL unreachable	Target URL is invalid or returns an error	Verify the URL is accessible