CLI15 Commands

CrawlForge CLI

CrawlForge ships a standalone CLI for one-off commands and scripting—no MCP client required. Install via npm, set an API key, and you're scraping in seconds.

Install & Authenticate

Install the CLI globally and authenticate by setting the CRAWLFORGE_API_KEY environment variable. Get a key from your API Keys dashboard.

Bash

Tip: Add the export line to your ~/.zshrc or ~/.bashrc so the key persists across shell sessions.

Commands

All 15 top-level commands. Run crawlforge <command> --help to see flags and options for any command.

Command	Purpose
`scrape`	Fetch and parse a single URL
`search`	Web search via Google or DuckDuckGo
`crawl`	Deep multi-page crawl with depth/filter controls
`map`	Site structure and link discovery
`extract`	Structured data extraction (CSS or LLM)
`track`	Set up content change tracking
`analyze`	Sentiment, language, readability analysis
`research`	Multi-stage deep research with source verification
`stealth`	Anti-detection scraping (Playwright or Camoufox)
`batch`	Bulk URL processing from a file or stdin
`actions`	Browser automation (clicks, scrolls, form fills)
`localize`	Geo-targeted scrape from 26 countries
`llmstxt`	Generate llms.txt and llms-full.txt
`template`	Run a pre-built site template (github-repo, amazon-product, etc.)
`monitor`	Run scheduled change monitoring

Examples

1. Scrape a single page with CSS selectors

Bash

Sample output:

Json

2. Run deep research with multiple sources

Bash

Crawls 10 sources, follows links 3 levels deep, verifies citations, and emits a structured research report. Costs 10 credits.

3. Use a pre-built site template

Bash

Templates extract a known schema for popular sites (GitHub repos, Amazon products, YouTube videos, and more) without writing selectors yourself.

Global Flags

These flags work with any command and shape how output is delivered.

Flag	Description
`--json`	Emit raw JSON to stdout (machine-readable)
`--output <file>`	Write results to a file instead of stdout
`--timeout <ms>`	Override the default request timeout
`--verbose`	Print debug logs and request details

Next Steps

API Reference

All 23 tools, parameters, and credit costs.

MCP Protocol

Use the same backend from any MCP-compatible client.

Getting Started

5-minute quickstart for new accounts.

Ready to scrape from the terminal?

Commands

All 15 top-level commands. Run crawlforge <command> --help to see flags and options for any command.

Command	Purpose
`scrape`	Fetch and parse a single URL
`search`	Web search via Google or DuckDuckGo
`crawl`	Deep multi-page crawl with depth/filter controls
`map`	Site structure and link discovery
`extract`	Structured data extraction (CSS or LLM)
`track`	Set up content change tracking
`analyze`	Sentiment, language, readability analysis
`research`	Multi-stage deep research with source verification
`stealth`	Anti-detection scraping (Playwright or Camoufox)
`batch`	Bulk URL processing from a file or stdin
`actions`	Browser automation (clicks, scrolls, form fills)
`localize`	Geo-targeted scrape from 26 countries
`llmstxt`	Generate llms.txt and llms-full.txt
`template`	Run a pre-built site template (github-repo, amazon-product, etc.)
`monitor`	Run scheduled change monitoring

Examples

1. Scrape a single page with CSS selectors

Bash

Sample output:

Json

2. Run deep research with multiple sources

Bash

Crawls 10 sources, follows links 3 levels deep, verifies citations, and emits a structured research report. Costs 10 credits.

3. Use a pre-built site template

Bash

Templates extract a known schema for popular sites (GitHub repos, Amazon products, YouTube videos, and more) without writing selectors yourself.

Flag

Description

--json

Emit raw JSON to stdout (machine-readable)

--output <file>

Write results to a file instead of stdout

--timeout <ms>

Override the default request timeout

--verbose

Print debug logs and request details