We ran our own form-label check on our own UI. It flagged one thing, and it was wrong.
We check our own code with the same rules we sell. This week we ran the form-label and focus check (rule 1026) on our UI. It is the kind of check that catches an input with no label, or a focus outline removed without a
We check our own code with the same rules we sell. This week we ran the form-label and focus check (rule 1026) on our UI. It is the kind of check that catches an input with no label, or a focus outline removed without a replacement.
What we ran
We looked at every component file in our app that removes the browser focus outline (outline-none). That is 53 files, scanned on 21 November 2026. The check looks at each element on its own. For a native input, textarea, select, button or link, it flags a removed outline with no replacement focus style. For an input, it flags a missing label unless one of these exists: a wrapping <label>, aria-label, aria-labelledby, title, or an id that a label points to.
What it found
One hit. It was in RuleKnowledgeBaseModal.tsx, which is the help text that explains each rule. The hit was an example of a form, written as text inside a string, not a live form on the page.
So the finding was wrong. The check read a code sample inside a string as if it were real markup.
Why it matters
A check that reads example text as real code will send people to fix things that are not broken. That costs time, and it makes people stop trusting the next alert. We found this in our own product before any customer saw it, which is the reason to run the check on yourself first.
What we are doing
We are deciding whether to make the check skip text inside strings and documentation examples. We will report the result here either way, with the before and after on the same 53 files.
If you run a similar check, ask one question about each hit: is this markup the browser will render, or text that only looks like markup? That question would have saved us this week.
Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.