generated: '2026-08-29'
method: searched
source: 'https://docs.crawl4ai.com/core/cli/'
provider: Crawl4AI
providerId: crawl4ai
name: crwl
description: >-
  Crawl4AI ships a first-party CLI, `crwl`, installed automatically with the
  Python library. It drives the local crawler — not the Cloud API — and covers
  crawling, extraction, content filtering and an LLM Q&A mode over crawled content.
install:
  - method: pip
    command: pip install crawl4ai
    note: 'The CLI "will be installed automatically when you install the library".'
  - method: setup
    command: crawl4ai-setup
    note: 'Post-install browser/environment setup, referenced by the plugin switch skill.'
binary: crwl
package: 'pypi:crawl4ai (see packages/crawl4ai-packages.yml)'
targets: local library
commands:
  - form: 'crwl <url>'
    description: Basic crawl of a single URL.
  - form: 'crwl <url> -o <format>'
    description: 'Choose the output format: all | json | markdown/md | markdown-fit/md-fit.'
  - form: 'crwl <url> -o json -v --bypass-cache'
    description: Verbose JSON output with the cache bypassed.
  - form: 'crwl --example'
    description: Print usage examples.
  - form: 'crwl <url> -B browser.yml | -b "headless=true,viewport_width=1280"'
    description: 'Browser configuration, by YAML file or inline key=value.'
  - form: 'crwl <url> -C crawler.yml | -c "css_selector=#main,scan_full_page=true"'
    description: 'Crawler configuration, by YAML file or inline key=value.'
  - form: 'crwl <url> -e extract_css.yml -s css_schema.json -o json'
    description: 'CSS/XPath (json-css) structured extraction against a schema file.'
  - form: 'crwl <url> -e extract_llm.yml -s llm_schema.json -o json'
    description: LLM-based structured extraction against a JSON Schema.
  - form: 'crwl <url> -f filter_bm25.yml -o markdown-fit'
    description: 'Content filtering — bm25 or pruning strategy with a query and threshold.'
  - form: 'crwl <url> -q "What is the main topic discussed?"'
    description: LLM Q&A over the crawled content.
flags:
  - {flag: '-o', name: output, values: [all, json, markdown, md, markdown-fit, md-fit]}
  - {flag: '-B', name: browser-config, type: yaml-file}
  - {flag: '-b', name: browser-params, type: inline-kv}
  - {flag: '-C', name: crawler-config, type: yaml-file}
  - {flag: '-c', name: crawler-params, type: inline-kv}
  - {flag: '-e', name: extraction-config, type: yaml-file}
  - {flag: '-s', name: schema, type: json-file}
  - {flag: '-f', name: filter-config, type: yaml-file}
  - {flag: '-q', name: question, type: string}
  - {flag: '-v', name: verbose}
  - {flag: '--bypass-cache'}
  - {flag: '--example'}
config:
  global_file: '~/.crawl4ai/global.yml'
  note: >-
    On first LLM use the CLI prompts for a provider and API token and stores them
    in the global config. Any LiteLLM provider is supported; Ollama needs no token.
cloud_cli:
  present: false
  note: >-
    There is no CLI for the hosted Cloud API. The nearest first-party equivalents
    are the Claude Code plugin scripts (crawl.py, poll-job.py, health.sh,
    test-key.sh) shipped in crawl4ai-cloud-sdk, and the documented raw curl calls.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com