WCAGpilot Worker Test Fixtures

Static good/bad pages for exercising every modular worker. Each worker has isolated single-check files plus an all-bad.html (and where useful all-good.html) that packs several issues onto one page.

Legend: PASS · FAIL · WARNING · N/A · AI outcome depends on the Gemini model and is not 100% deterministic.

How to test

HTML-parsing workers accept the page markup directly in the request body, e.g.:

POST https://api.wcagpilot.com/w/headings with { "html": "<the page source>" }

URL-fetching workers (axe, keyboard-focus, reflow) need a live public URL — deploy this folder, then pass the deployed page URL, e.g. { "pages": [{ "url": "https://YOUR-HOST/axe/all-bad.html" }] }.

Headings — /w/headings

Dynamic rendering — shadow DOM & client-rendered SPA

These pages have no headings in their static HTML — the outline lives in shadow roots (web components) or is painted by JS after load. A raw { "html": ... } parse sees an empty page (false WARNING); the extension's in-page extraction and headless-render workers (axe) see the real outline. Use these to verify the analyzer reaches rendered content.

Page titles — /w/titles AI (fully model-driven)

Skip links — /w/skip-links

Forms — /w/forms (label quality is AI)

Media — /w/media

Keyboard focus — /w/keyboard-focus (needs public URL)

Reflow & zoom — /w/reflow (needs public URL)

Alt text — /w/alt-texts-v2 (quality is AI vision)

Axe-core — /w/axe (needs public URL, headless render)

Accessibility statement — /w/statement/find + /w/statement/parse

The discovery worker crawls a homepage for a statement link; the link below gives it a fully parseable page (conformance status, dates, evaluation method, known issues, technologies, contact, enforcement).

• Accessibility Statement — parseable

Page discovery — /w/pages

This index links every fixture above, so the AI page-discovery worker can use it as a homepage to crawl and rank candidate pages.