Scanning and Fixing

Scanning

Every page is checked against a defined set of accessibility rules based on the WCAG 2.1 AA standard — covering things like whether images have meaningful alt text, whether headings and page structure are logical, whether color contrast is strong enough to read, whether everything can be operated with a keyboard alone (not just a mouse), whether interactive elements show a visible focus state, whether the page's language is declared, and whether form fields have proper labels.

Each issue found is classified by severity — critical, serious, warning, or notice — reflecting how much it actually blocks someone from using the page versus a minor best-practice gap.

Scoring

A page starts at a perfect score and loses points based on the severity of what's found (bigger deductions for critical issues than minor ones). That score maps to a letter grade (A through F), giving a quick read on a page's overall accessibility health without having to read every individual issue.

Grouping by impact (POUR)

Beyond a raw list, issues are organized into four categories that describe what kind of barrier they create:

• Perceivable — can people perceive the content at all (images, contrast, captions)?
• Operable — can people navigate and interact (keyboard access, focus, timing)?
• Understandable — is the content and behavior predictable and clear (labels, language, instructions)?
• Robust — does it work reliably across assistive technologies (valid structure, ARIA)?

This helps prioritize fixes by the type of user impact, not just a flat severity list.

Fixing

For each issue, there's an AI-assisted path to resolve it: generating appropriate alt text for an image, suggesting a code-level fix for a structural problem, or explaining in plain language why something fails and what to change. Rather than just flagging a problem, the goal is to get from "here's what's wrong" to "here's a specific fix" with minimal manual digging through WCAG documentation.

Rescanning

Because fixes are applied back to the actual page content, a follow-up scan is how you confirm an issue is resolved — the score and grade update to reflect the real current state of the page rather than a one-time snapshot.