How It Works

axe-core is the industry-standard open-source WCAG rule engine, maintained by Deque Systems. It is used by Microsoft, Google, and thousands of organisations worldwide to test web accessibility.

• Over 300 rules covering WCAG 2.1 and 2.2 Level A and AA success criteria
• Designed for high accuracy -- every issue reported maps to a real WCAG criterion
• Tests for colour contrast, missing alt text, ARIA misuse, form labels, heading structure, landmark roles, and much more
• Each issue includes the exact element, the rule violated, and a clear explanation of how to fix it

Many accessibility scanners only analyse static HTML. That misses issues in JavaScript-rendered content, single-page applications, and dynamic interfaces. Lumi uses Playwright to run a full headless Chromium browser for every scan.

• Full JavaScript execution -- React, Next.js, Vue, Angular, and other framework-built sites are tested as users actually experience them
• Waits for dynamic content to load before analysing the page
• Captures the rendered DOM, computed styles, and visual layout
• Handles authentication walls, cookie banners, and other interactive elements

Keyboard navigation is one of the most critical aspects of web accessibility, and one of the hardest to test automatically. We have built custom testing logic that goes beyond what standard rule engines check.

• Focus not obscured (WCAG 2.2): Detects focused elements fully hidden behind sticky headers, cookie banners, or fixed footers — a common issue no other tool tests automatically
• Keyboard traps: Detects situations where a user can tab into an element (such as a modal or widget) but cannot tab out again
• Focus visibility: Checks that focused elements have a visible indicator so keyboard users can see where they are on the page
• Skip links: Checks for the presence and functionality of skip navigation links that let users bypass repeated content
• Tab order: Verifies that interactive elements follow a logical sequence and that tabindex is used correctly
• Keyboard activation: Detects interactive elements with click handlers or ARIA roles that are unreachable by keyboard

Our second scanning engine, HTML_CodeSniffer runs alongside axe-core to catch issues that a single engine might miss. Results are deduplicated -- you see one unified report, not duplicate noise.

• Checks WCAG 2.1 Level AA with a different rule interpretation, catching edge cases axe-core classifies differently
• Particularly strong on content structure, link purpose, and language attribute checks
• Issues are tagged with their source engine for full transparency
• Deduplicated against axe-core results so the same element is never reported twice for the same WCAG criterion

What automated scanning can and cannot do

We believe in transparency. Most single-engine accessibility scanners reliably detect around 30% of WCAG 2.2 issues. Lumi's multi-engine approach — combining axe-core, HTML_CodeSniffer, keyboard testing, and our proprietary engine — pushes that closer to 50%. These tend to be the most common and impactful issues: missing alt text, colour contrast failures, broken ARIA, missing form labels, structural problems, and visual layout issues.

The remaining issues require human judgement. Things like whether content is logically ordered for screen reader users, whether interactive patterns are intuitive, whether error messages are helpful, and whether the overall experience is genuinely usable for people with disabilities. No tool can fully replace a human tester.

Automated scanning is the essential first step. It catches the most common issues quickly and affordably, and gives you a baseline to track over time. But if you need full WCAG compliance - for legal requirements, procurement, or because you genuinely want your product to be accessible to everyone - you will need expert human review as well.