CLI & Skill

Firecrawl CLI and AI agent skill for seamless web scraping integration

The Firecrawl CLI provides command-line access to the Firecrawl API, while the Firecrawl Skill enables AI agents like Claude Code, Codex, and OpenCode to use Firecrawl automatically.

Installation

Install the Firecrawl skill to enable CLI and AI agent integration:

npx skills add firecrawl/cli

After installation, restart your AI agent to activate the skill.

What is the Firecrawl Skill?

The Firecrawl skill is a tool that allows AI coding agents to:

• Automatically scrape web pages when you ask questions about websites
• Crawl entire sites to gather context for development tasks
• Search the web for current information
• Extract structured data from pages
• Use the Agent feature for autonomous web research

Once installed, you can simply ask your AI agent to “scrape the documentation from example.com” or “find the latest pricing from their website”, and it will use Firecrawl automatically.

CLI Usage

The CLI is also available directly after installing the skill.

Authentication

Set your API key as an environment variable:

export FIRECRAWL_API_KEY="fc-YOUR_API_KEY"

Get your API key from firecrawl.dev.

Scrape a URL

firecrawl scrape https://firecrawl.dev

With options:

firecrawl scrape https://firecrawl.dev \
  --formats markdown,html \
  --only-main-content

Crawl a Website

Crawl with automatic wait for completion:

firecrawl crawl https://firecrawl.dev --limit 50

With advanced options:

firecrawl crawl https://firecrawl.dev \
  --limit 100 \
  --exclude-paths "blog/*,admin/*" \
  --formats markdown

Map a Website

Discover all URLs:

firecrawl map https://firecrawl.dev

Search for specific content:

firecrawl map https://firecrawl.dev --search pricing

Search the Web

firecrawl search "firecrawl web scraping" --limit 10

Use the Agent

Autonomous web research:

firecrawl agent "Find the pricing plans for Notion"

With specific URLs:

firecrawl agent "Compare the features" \
  --urls https://notion.so,https://airtable.com

Batch Scrape

Scrape multiple URLs:

firecrawl batch-scrape \
  https://firecrawl.dev \
  https://docs.firecrawl.dev \
  --formats markdown

Using with AI Agents

Once the skill is installed, you can interact with your AI agent naturally:

Example Prompts

Scraping:

Scrape the documentation from https://docs.firecrawl.dev and summarize the main features

Crawling:

Crawl the Firecrawl blog and extract all article titles and URLs

Agent:

Find the latest pricing information for Notion, Airtable, and Coda, then compare them

Search:

Search for the best web scraping tools in 2024 and give me a summary

How It Works

When you ask your AI agent to perform web scraping tasks:

1. The agent recognizes it needs web data
2. It automatically invokes the Firecrawl skill
3. Firecrawl fetches the content using your API key
4. The agent receives clean, LLM-ready data
5. The agent uses that data to answer your question

No manual intervention required!

Supported Agents

The Firecrawl skill works with:

• Claude Code (Anthropic)
• OpenCode (open-source coding agent)
• Codex (GitHub Copilot)
• Any agent that supports the Skills protocol

Configuration

Environment Variables

# Required: Your Firecrawl API key
export FIRECRAWL_API_KEY="fc-YOUR_API_KEY"

# Optional: Custom API URL (for self-hosted)
export FIRECRAWL_API_URL="https://api.firecrawl.dev"

Skill Configuration

The skill can be configured via ~/.config/firecrawl/config.json:

{
  "apiKey": "fc-YOUR_API_KEY",
  "apiUrl": "https://api.firecrawl.dev",
  "defaultFormats": ["markdown"],
  "maxCrawlLimit": 100
}

CLI Reference

Global Flags

• --api-key - Your Firecrawl API key (overrides environment variable)
• --api-url - Custom API URL
• --output - Output file path (default: stdout)
• --format - Output format: json, markdown, text
• --quiet - Suppress progress messages
• --help - Show help

Commands

scrape <url>

Scrape a single URL.

Options:

• --formats - Output formats (comma-separated): markdown, html, screenshot, links, json
• --only-main-content - Extract only main content
• --include-tags - HTML tags to include
• --exclude-tags - HTML tags to exclude
• --wait-for - Wait time in milliseconds

crawl <url>

Crawl a website.

Options:

• --limit - Max pages to crawl
• --max-depth - Maximum discovery depth
• --exclude-paths - Paths to exclude (comma-separated)
• --include-paths - Paths to include (comma-separated)
• --formats - Output formats
• --poll-interval - Polling interval in seconds

map <url>

Map all URLs on a website.

Options:

• --search - Search query to filter URLs
• --limit - Max URLs to return
• --include-subdomains - Include subdomains

search <query>

Search the web.

Options:

• --limit - Number of results
• --scrape - Also scrape the results
• --formats - Output formats for scraped content

agent <prompt>

Run the AI agent.

Options:

• --urls - URLs to focus on (comma-separated)
• --model - Model to use: spark-1-mini, spark-1-pro
• --schema - JSON schema file for structured output

batch-scrape <urls...>

Scrape multiple URLs.

Options:

• --formats - Output formats
• --poll-interval - Polling interval in seconds

Output Formats

The CLI supports multiple output formats:

JSON (default)

Structured JSON output:

firecrawl scrape https://firecrawl.dev --format json

Markdown

Clean markdown output:

firecrawl scrape https://firecrawl.dev --format markdown

Text

Plain text output:

firecrawl scrape https://firecrawl.dev --format text

Error Handling

The CLI provides clear error messages:

# Missing API key
Error: FIRECRAWL_API_KEY environment variable not set

# Invalid URL
Error: Invalid URL format

# Rate limit
Error: Rate limit exceeded. Please try again later.

# Authentication failed
Error: Invalid API key

Examples

Scrape and Save

firecrawl scrape https://firecrawl.dev \
  --formats markdown \
  --output firecrawl.md

Crawl with Filtering

firecrawl crawl https://docs.firecrawl.dev \
  --limit 50 \
  --include-paths "docs/*" \
  --exclude-paths "blog/*" \
  --formats markdown,html \
  --output crawl-results.json

Agent Research

firecrawl agent "Find the top 3 features of Firecrawl and their pricing" \
  --model spark-1-pro \
  --output research.json

Batch Processing

cat urls.txt | xargs firecrawl batch-scrape \
  --formats markdown \
  --output batch-results.json

Updating

Update to the latest version:

npx skills update firecrawl/cli

Uninstalling

Remove the skill:

npx skills remove firecrawl/cli

Resources

• Skills Protocol
• GitHub Repository
• API Reference
• Firecrawl Documentation

Support

For issues or questions:

• GitHub Issues
• Discord Community
• Documentation