generated: '2026-08-29'
method: searched
source: >-
  https://gate.crawl4ai.com/ (playground + "Get a key"),
  https://docs.crawl4ai.com/core/self-hosting/, https://gate.crawl4ai.com/health
provider: Crawl4AI
providerId: crawl4ai
description: >-
  Crawl4AI's try-before-you-buy surface is an in-page playground on the Cloud API
  landing page plus an instantly-issued 24-hour key that needs no signup. There
  is no separate sandbox HOST and no test/live key split — the playground and the
  free key run against the same production API on real URLs.
sandbox_host: none
test_mode: none
key_split:
  present: false
  note: >-
    One key prefix, sk_live_, for every tier. No sk_test_ equivalent, so there is
    no way to point an integration at a non-billing environment.
surfaces:
  - name: Cloud API playground
    url: https://gate.crawl4ai.com/#playground
    type: hosted-console
    auth_required: false
    note: >-
      Interactive runner on the landing page for the scrape, extract, search and
      answer endpoints — "Type a URL above and hit Run - real output renders here",
      with a "Copy full command" affordance that emits the equivalent curl. The
      page labels it "Live · play limits", i.e. it calls production under a
      reduced allowance. Requires JavaScript; the page says so itself and points
      non-JS readers at docs.crawl4ai.com.
  - name: Play pass (24-hour key)
    url: https://gate.crawl4ai.com/
    type: instant-credential
    auth_required: false
    ttl: 24 hours
    note: >-
      "Get a free key: visit https://gate.crawl4ai.com/ and click 'Get a key' (a
      24-hour key is issued instantly; verify your email to keep it)." The Terms
      name it as a distinct tier: "a free Community tier, a 24-hour Play pass, and
      a paid Supporter subscription."
  - name: Anonymous health check
    url: https://gate.crawl4ai.com/health
    type: unauthenticated-endpoint
    auth_required: false
    probed: {status: 200, body: '{"ok":true}', date: '2026-08-29'}
  - name: Anonymous MCP introspection
    url: https://gate.crawl4ai.com/mcp
    type: unauthenticated-endpoint
    auth_required: false
    note: >-
      tools/list answers without a credential, so an agent can discover the full
      tool surface and its input schemas before anyone signs up. tools/call is gated.
  - name: Self-hosted playground
    url: 'http://localhost:11235/playground'
    type: self-hosted-console
    auth_required: 'from 0.9.0, yes — the server binds loopback and requires a token'
    note: >-
      "Visit /playground to access the interactive testing interface" once the
      Docker image is running. The whole self-hosted server is in effect a free,
      unmetered sandbox for the crawler itself.
  - name: Documentation LLM context builder
    url: https://docs.crawl4ai.com/
    type: docs-tool
    note: >-
      The docs site ships an "Ask AI" assistant and an "LLM Context Builder" for
      assembling a context bundle from the documentation.
test_fixtures:
  published: false
  note: >-
    No test URLs, magic identifiers, fixture pages or trigger values are published
    — appropriate for a crawler, whose input is the live public web. The docs'
    own worked examples use https://example.com and https://news.ycombinator.com.
  never_invent: >-
    No test credentials or fixture values were authored for this file; Crawl4AI
    publishes none.
time_simulation:
  present: false
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com