AI fixes you
actually approve.
Open any failing rule and you see one panel: the compliance requirement, the intent behind it, the AI's suggested fix, and a plain-English description of what changed. Review it in one go. Revise or approve.
Everything you need to decide, in one panel.
No clicking through tabs. No bouncing between requirement, code, and explanation. Open an issue and the four pieces of context that matter are all on screen at once.
Requirement
The exact WCAG rule, its level (A, AA, AAA), and a plain-English summary of what it asks for.
Intent of the rule
Why the rule exists. Who it protects. What gets lost when the rule isn't met. The "why" before the "what".
What the AI changed and why
The AI's reasoning in plain language — what it decided to do, what alternatives it considered, why it picked this approach for your context.
Suggested change
The actual HTML diff. Before on top, after on bottom, the new bits highlighted. No hidden edits.
Image is missing alt text
Requirement
Every <img> element that conveys meaning must have a non-empty alt attribute describing its purpose. Decorative images should use alt="" so assistive tech can skip them.
Intent of the rule
Screen readers can't see images. Without an alt attribute, blind and low-vision users get silence where sighted users see a picture. The rule exists so meaning carried by an image still reaches everyone — and so purely decorative images don't waste a screen-reader user's attention.
What the AI changed and why
I added a descriptive alt attribute. The image is editorial — it's the leadership photo on the About page — so I used a meaningful description instead of alt="". The phrasing matches the surrounding paragraph so the page reads naturally on a screen reader.
Suggested change
<img src="/team-photo.jpg" class="rounded-lg" />
<img src="/team-photo.jpg" class="rounded-lg" alt="Acme leadership team in front of the new SF office, March 2026" />
Revise it, or approve it.
That's the whole flow. Every revision and every approval is logged to an immutable audit trail.
Request changes
Type what you want different — in plain language. Examples:
- "Use a button instead of a div with role=button."
- "Match the existing class naming on this template."
- "Don't add ARIA where semantic HTML works."
The AI revises and shows you a fresh panel — same four sections, new suggestion. Iterate as many times as you need.
Approve fix
You click approve. The fix runs through WordPress's built-in content security check, then applies to your post content (or via the runtime layer for template-level fixes).
- Sanitized before it touches your site
- Logged to the immutable audit trail
- Reversible — fixes are permanent only because you approved them
Auto-fix is how sites get sued.
Accessibility overlay widgets — the ones that promise to "fix your site instantly with AI" — have been named in hundreds of ADA lawsuits. They patch the DOM at runtime, fail in edge cases, and break assistive technology. Courts have consistently ruled that they don't constitute compliance.
We do the opposite. The AI proposes. You review the panel. You decide whether it fits your context, your codebase, and your design system. When you approve, the fix applies to your real HTML — sanitized, audited, reversible.
- Patch your DOM at runtime with a JavaScript overlay
- Auto-apply fixes without your explicit approval
- Hide what changed in a diff you can't read
- Send any AI-generated HTML to your site without sanitization
One workflow. Four parts that fit together.
Review the fixes. Approve what fits.
The AI suggests, you decide. Every approved change is sanitized, audited, and reversible. Nothing ships without your sign-off.
Free scan covers one URL. The full plugin scans every page and suggests AI fixes you review.