Firecrawl CLI

When using firecrawl setup commands, you can target specific AI coding agents using the --agent flag. If no flag is provided, the CLI attempts to install to all detected harnesses.

| Harness      | `--agent` value  |
| ------------ | ---------------- |
| Claude Code | `claude-code`    |
| Codex        | `codex`          |
| Cursor       | `cursor`         |
| Windsurf     | `windsurf`       |
| OpenCode     | `opencode`       |
| OpenClaw     | `openclaw`       |
| OpenHands    | `openhands`      |
| Hermes Agent | `hermes-agent`   |

Firecrawl monitors detect content changes on websites and provide notifications via webhooks or email. They eliminate the need for manual cron jobs, scrapers, or diffing scripts.

Monitors are divided into two main types:

1. Page Monitors: Watch specific URLs (single pages, batches of URLs, or entire sites via crawling) for changes.
2. Web Monitors: Watch the entire web for new results matching a specific goal using search queries. These do not watch a fixed URL but alert you when new, relevant content is discovered via search.

Each page in a check is assigned a status: same, new, changed, removed, or error. An AI judge uses the provided --goal to filter out noise like whitespace, formatting, or tracking parameters, ensuring notifications only trigger on substantive content changes.

To get the most out of the firecrawl agent, follow these tips:

• Use --wait: Always include the --wait flag to receive results directly in your terminal/output. If you omit it, the command will only return a job ID.
• Use --schema: For predictable, structured output, always provide a schema. Without it, the agent returns freeform data.
• Manage costs: Agent runs consume more credits than simple scrapes. Use --max-credits <n> to cap your spending for a specific run.
• Choose the right tool: For simple single-page extraction, prefer scrape—it's faster and cheaper.

A common and effective pattern is to use map --search to identify the exact URL you need, and then use the scrape command on that specific URL. This avoids wasting resources scraping irrelevant pages.

Example Workflow:

1. Run firecrawl map https://docs.example.com --search "auth".
2. Identify the correct URL from the results (e.g., /docs/api/authentication).
3. Run firecrawl scrape on that specific URL.

# Step 1: Find the URL
firecrawl map https://docs.example.com --search "auth"

# Step 2: Scrape the discovered URL
firecrawl scrape https://docs.example.com/docs/api/authentication

Firecrawl allows you to direct output to different destinations and formats:

• Stdout (Default): Output is printed directly to the terminal.
• Save to File: Use the -o <filename> flag to save output to a file.
• JSON/Multiple Formats:
  • If you specify a single format (e.g., --format markdown), the CLI outputs the raw content.
  • If you specify multiple formats (comma-separated, e.g., --format markdown,links), the CLI outputs a JSON object containing all requested data.
• Pretty Print: Use the --pretty flag with JSON output for better readability.

# Output to stdout (default)
firecrawl https://example.com

# Pipe to another command
firecrawl https://example.com | head -50

# Save to file
firecrawl https://example.com -o output.md

# JSON output with multiple formats
firecrawl https://example.com --format markdown,links,images

Once installed, verify your configuration by performing a small test request. It is recommended to output the result to a file to ensure write permissions and output handling are working correctly.

Run the following command to scrape the Firecrawl website into a local directory:

mkdir -p .firecrawl
firecrawl scrape "https://firecrawl.dev" -o .firecrawl/install-check.md

The firecrawl download command is an experimental convenience tool that combines map and scrape operations to save an entire website as local files. It first maps the site to discover all pages and then scrapes each page into organized, nested directories under .firecrawl/.

Note: Always pass the -y flag to skip the confirmation prompt, especially in automated workflows or scripts.

firecrawl download https://docs.example.com -y

The firecrawl map command is used to discover and list URLs on a website. It is particularly useful for finding specific subpages on large sites or obtaining a full list of URLs before performing a scrape or crawl.

Common use cases include:

• Finding a specific page on a large site using the --search filter.
• Listing all URLs on a site to understand its structure.
• Preparing for a workflow escalation pattern: search → scrape → map → crawl → interact.

# Find a specific page on a large site
firecrawl map "<url>" --search "authentication" -o .firecrawl/filtered.txt

# Get all URLs
firecrawl map "<url>" --limit 500 --json -o .firecrawl/urls.json

You can make Firecrawl the default web provider for supported AI agents. This disables the agent's native web fetch/search capabilities and routes all web work through Firecrawl.

Use firecrawl setup defaults to apply this. You can use -y to apply to all harnesses or --agent <agent> to target a specific one. Use --undo to restore original settings.

# Make Firecrawl default for all harnesses (non-interactive)
firecrawl setup defaults -y

# Make Firecrawl default for only Codex
firecrawl setup defaults --agent codex

# Undo changes for a specific agent
firecrawl setup defaults --undo --agent claude

You can run multiple independent Firecrawl operations in parallel using shell backgrounding. Before doing so, check your concurrency limit using firecrawl --status.

# Run multiple scrapes in parallel and wait for completion
firecrawl scrape "<url-1>" -o .firecrawl/1.md &
firecrawl scrape "<url-2>" -o .firecrawl/2.md &
firecrawl scrape "<url-3>" -o .firecrawl/3.md &
wait