Firecrawl Agent Skill

This skill enables autonomous deep web research using Firecrawl's

/agent

When to Use This Skill

Use this skill when you need to:

• Research companies - Find founders, funding, employee counts, tech stacks
• Gather competitive intelligence - Compare products, pricing, features
• Extract structured data - Get specific information in a defined schema
• Answer questions requiring web research - When information isn't in your knowledge base

Quick Start

Run a simple research query:

cd firecrawl-agent/scripts
python firecrawl_agent.py "Find the founders and founding year of Anthropic"

Prerequisites

1. Install dependencies:

   pip install -r scripts/requirements.txt
   

2. Set API key:

   export FIRECRAWL_API_KEY=your_api_key_here
   

   Get your API key at: https://www.firecrawl.dev/

Usage

Basic Research (No Schema)

python scripts/firecrawl_agent.py "What are the main features of Notion?"

Research with Structured Output

For predictable, structured responses, provide a JSON schema:

python scripts/firecrawl_agent.py \
  "Find information about Stripe" \
  --schema '{"company_name": "string", "founded_year": "number", "founders": ["string"], "headquarters": "string"}'

Command Line Options

prompt

--schema

--model

spark-1-mini

spark-1-pro

spark-1-mini

--urls

--max-credits

Model Selection

• spark-1-mini

  (default): Faster, cheaper, good for straightforward queries

• spark-1-pro

  : More capable, better for complex research requiring deeper navigation

# Use pro model for complex research
python scripts/firecrawl_agent.py \
  "Compare the pricing tiers of Notion, Coda, and Obsidian" \
  --model spark-1-pro

Providing Starting URLs

If you know relevant URLs, provide them to focus the search:

python scripts/firecrawl_agent.py \
  "Extract the pricing information" \
  --urls "https://stripe.com/pricing,https://stripe.com/enterprise"

Common Use Cases

Company Research

python scripts/firecrawl_agent.py \
  "Research Anthropic: founders, funding rounds, key products, and employee count" \
  --schema '{"name": "string", "founders": ["string"], "funding_total": "string", "products": ["string"], "employee_count": "string"}'

Product Comparison

python scripts/firecrawl_agent.py \
  "Compare Vercel and Netlify deployment platforms" \
  --model spark-1-pro

Contact Information

python scripts/firecrawl_agent.py \
  "Find contact information for Acme Corp" \
  --schema '{"email": "string", "phone": "string", "address": "string", "social_links": ["string"]}'

Output Format

The script outputs JSON with the following structure:

{
  "success": true,
  "status": "completed",
  "data": {
    // Your extracted data here
  },
  "sources": [
    "https://example.com/page1",
    "https://example.com/page2"
  ],
  "credits_used": 12
}

Error Handling

Common errors and solutions:

FIRECRAWL_API_KEY not set

Rate limit exceeded

Credit limit reached

maxCredits

--max-credits

Invalid schema

Cost Management

• The agent uses credits based on complexity and pages visited
• Free tier: 5 runs per day
• Set

  --max-credits

  to cap spending:

  python scripts/firecrawl_agent.py "Research topic" --max-credits 25
  

Reference Documentation

• See

  references/REFERENCE.md

  for complete API parameter documentation
• See

  references/SCHEMAS.md

  for common schema patterns
• See

  assets/example_schemas/

  for ready-to-use Pydantic models

Notes

• Firecrawl agent is in "research preview" - pricing is dynamic
• Results are available for 24 hours after completion
• Complex queries may take 30-60 seconds to complete