Can everyone use this page?

Lighthouse is Google's open source page auditor, the engine behind PageSpeed Insights. Its accessibility category runs a set of automated checks on one loaded page, such as image alt text, form labels, color contrast, and valid ARIA, and combines them into a number from 0 to 100. Each check has a weight, so a failed contrast check costs more points than a skipped heading level. The scoring page lists the weights.

Lighthouse colors 90 and above green, and IndieA11y follows that line. A page with no failed automated check passes, whatever else remains for a person to review. A score of 90 or more with failures reads as a few checks need work. Under 90 reads as needs attention.

The score is a summary, not the finding. The list of failed checks under it is what you act on. One heavy failure on a simple page can pull the number down fast, and a page with many small failures can still score in the 80s, so read the list before you read the number.

Every failed check in the result says in plain words what went wrong, why it matters to someone using assistive technology, and what to change, with up to three affected elements shown by selector and snippet and a link to the full Lighthouse guidance. The common fixes are short: write alt text, connect a label to each field, give every link and button a name, raise the text contrast, and set lang on the html element.

Use Copy issue beside one failed check to keep its examples, suggested fix, measured device, original timestamp, run warnings, and source in a task or an agent conversation. Deploy the fix and run the check again. A successful result is cached for six hours per device, so a re-run inside that window shows the same measurement and the same timestamp. The result names when it was measured, so you always know whether you are looking at the old page or the new one.

Each run measures one device. Mobile is the default and loads the page in a narrow viewport with a phone profile; desktop uses a wide one. Responsive layouts show different elements at each size, so a hidden menu, a smaller font, or a different background can fail on one device and pass on the other. Choose the other device on the form to run it separately. The result always names the device it measured and never mixes the two.

No. Automated checks find only the problems a machine can detect on one page in one run. They cannot tell whether alt text is meaningful, whether the page works by keyboard alone, whether a screen reader reads it in a sensible order, or whether the content makes sense. The result lists these manual checks beside the score, and they stay visible at 100.

The W3C says the same about every evaluation tool: automated checks find some problems, and people find the rest. A pass here is a good start on one page of one site, not WCAG conformance, not legal compliance, and not proof that a person who relies on assistive technology can use the product.

Yes. No account, no card, no premium plan, no upgrade nag, no hidden paywall. IndieA11y is free, all of it.

There is no public lookup history. A network check sends the public input needed for that check to its named source. Local file tools process file contents in your browser. Cookieless analytics counts visits and fixed actions, never entered values or file contents. Anything the tool saves for you stays in your browser.