ScanA11yWCAG 2.2 / A / AA / AAA

Accessibility,audited. Proofattached.

Site, Figma file, screenshot, PDF, Android or iOS build: you get the element that failed, the criterion it breaks, and the fix.

evidence / page 1 of 5serious

1.4.3Contrast (Minimum)AA

2.94:1measured4.5:1 required

Foreground #9aa0a6 on #ffffff. Darkening to #6b7075 reaches 4.83:1.

One finding, as the workbench renders it.

Start an audit

Five a day without an account. Everything deletes itself in 24 hours.

Evidence, not a rule name: The crop from the scan, the criterion (1.4.3 Contrast (Minimum), AA), the measured ratio against the one required, and the fix. No crop beats a wrong crop.
A matrix across all 87 criteria: Passed, failed, needs review, not tested. Nothing untested is ever reported as passed.
A 16:9 PDF, a VPAT 2.5, a maturity score: One finding per printed page with its screenshot, scored across governance, design, dev, QA and ops.

compliance matrix / 87 criteria

URL

A live site

Crawls up to 5 pages: axe-core 4.13, keyboard order, reflow at 375 and 768, evidence crops.

FIG

A Figma file

Frames over the REST API: contrast pairs, touch targets, images with no description. Connect an account or paste a link.

IMG

A UI screenshot

Element detection first, then vision-model suggestions that stay advisory.

PDF

A PDF document

PDF/UA structure up to 25MB: tags, reading order, language, alt text, and scans posing as documents.

APK

An Android package

aapt2 manifest analysis, plus live checks for labels, target size and contrast when an emulator is running.

IPA

An iOS bundle

Info.plist and asset inspection, plus a guided VoiceOver checklist for the dynamic half.

Three modes over one audit: the page on one side, the WCAG 2.2 checklist on the other.

Checklist: Every criterion by principle, with findings, evidence and verdict, filterable to what each module covered.
Inspect: Click any element for its contrast, target size and accessible name; simulate colour vision deficiency, larger text, reduced motion.
Screen reader: A transcript built from the accessibility tree, read aloud in the browser, nothing to install.

Same axe-core engine as axe DevTools, Lighthouse and Accessibility Insights. What changes is how many tools you need around it.

ScanA11y compared with common accessibility testing tools, by type, cost and coverage.
Tool	Type	Cost	Coverage
ScanA11y	Automated plus guided manual	Free, open source (MIT)	URL, Figma, screenshot, PDF, APK, IPA
axe DevTools (Deque)	Browser extension, paid platform	Free extension, Pro from ~$1,140/seat/yr	URL only, same axe-core engine
WAVE (WebAIM)	Browser extension, API	Free extension, paid API	URL only, visual overlay
Google Lighthouse	Built into Chrome DevTools	Free	URL only, automated checks
JAWS	Screen reader, manual testing	~$95/yr to $1,000 one-time	Manual only, Windows
NVDA	Screen reader, manual testing	Free	Manual only, Windows
Level Access, Deque platform	Enterprise SaaS suite	$500 to $8,000/seat/yr	Automated plus manual, usually one input type

Published list prices, which move; the column that matters is coverage.

Automated testing cannot certify conformance.: It catches a subset. The rest becomes a guided checklist for a human, never a silent pass.
Vision models never create a finding.: They suggest, into needs review, labelled advisory. Contrast is deterministic colour maths.
iOS dynamic testing needs macOS.: Nothing on Windows or Linux can drive the iOS Simulator. You get static bundle analysis and a Mac checklist.
Sites behind a bot wall will not scan.: The challenge page is reported as blocked, with the reason, not as a clean pass.

MIT licensed, with every engine in the audit path open source. Disagree with a finding and you can read the check that produced it.

A personal portfolio project, not affiliated with any employer. Built on Next.js, Supabase, Inngest, Playwright and axe-core.