Public Accessibility Snapshot Agent

Run a lightweight accessibility preflight on public HTML pages. The Actor checks visible page evidence for language declaration, title, heading structure, image alt text, form labels, ambiguous link/button text, ARIA roles, landmarks, evidence snippets, and a stable change hash.

This is designed for AI agents, website QA, growth teams, content operations, and procurement workflows that need a quick public-page risk snapshot before spending time on a full manual accessibility audit.

What It Returns

• accessibilityScore: a bounded basic-risk score from visible HTML evidence.
• riskLabels: signals such as missing_language, missing_h1, images_missing_alt, unlabeled_form_controls, and ambiguous_link_or_button_text.
• languageSignal, headingOutline, imageAltSummary, formControlSummary, ambiguousTextSummary, and ariaLandmarkSummary.
• evidenceSnippets and evidenceUrls so an AI agent can explain the result.
• accessibilityHash and changeStatus for change monitoring.

Good Fits

• Batch preflight of public marketing pages before a redesign or migration.
• AI-agent website QA where you need structured accessibility risk labels.
• Monitoring whether basic public-page accessibility signals changed.
• Vendor or competitor website review where only public page evidence is used.

Safety Boundary

• Public HTTP and HTTPS pages only.
• No browser rendering, JavaScript execution, login, cookies, form submission, CAPTCHA bypassing, external resource downloading, or screen-reader emulation.
• Query strings, fragments, URL credentials, path parameters, private networks, localhost, .local, account-like sensitive paths, and token-like path segments are rejected or safely redacted.
• HTML/text responses are capped at 3 MB.
• This Actor is not a WCAG certification, legal compliance opinion, or substitute for a full manual audit.

Pricing Behavior

The pay-per-event setup charges only for useful public accessibility snapshots:

• apify-actor-start: charged once when a run starts.
• useful-accessibility-snapshot-result: charged only for a new or changed snapshot with enough visible evidence and any required focus-check match.

The Actor intentionally does not configure apify-default-dataset-item, so errors, duplicate inputs, unchanged snapshots, focus misses, and low-evidence records can be written without charging the useful-result event.

Example Input

{
  "startUrls": ["https://www.w3.org/WAI/"],
  "focusChecks": ["headings", "alt", "forms"],
  "requestTimeoutSecs": 20
}