Strict code review that ignores style
codingReviews a diff for correctness, edge cases, and regressions — and refuses to comment on formatting.
Checks markup against the failures screen readers actually hit, not a checklist of rules.
Before shipping a component that someone will use with a keyboard or a screen reader.
Audit this markup for accessibility defects that a real user would hit.
Report, in order of impact:
1. Keyboard traps and focus order problems.
2. Missing or wrong accessible names (buttons, inputs, landmarks).
3. State that is visual only (aria-expanded, aria-selected, aria-live missing).
4. Colour and contrast issues you can infer from the classes.
5. Anything that will read as noise to a screen reader.
Rules:
- For each issue: the element, what a keyboard or screen-reader user experiences, and the minimal fix.
- Do not suggest adding ARIA where a native element would do. That is the most common wrong fix.
- If the markup is fine, say so.
Markup:
<html>
Say what the component is supposed to do — a dialog and a tooltip need different audits.
The prohibition on gratuitous ARIA is what separates a useful audit from a linter. Wrong ARIA is worse than none, and models default to adding attributes.
Reviews a diff for correctness, edge cases, and regressions — and refuses to comment on formatting.
Extracts or restructures code while proving the public behavior stays identical.
Writes the failing test first, then the fix, and refuses to skip the reproduction.