Example threads. These questions and answers were written by the SDODS team to document real problems and how to fix them. The people named are illustrative, not real users. Questions asked on this site are marked new.
We accepted that the tag does nothing and we are adding our own step, so we will shortly have a machine-readable list of violations per page.
The advice I keep getting is "do not gate on it directly, triage first". That sounds like the advice that turns a check into a report nobody reads. What specifically goes wrong if we just fail the build on anything serious or above?
I audit these for a living, so: what goes wrong is that a meaningful fraction of the findings are true statements about the DOM and wrong statements about the product.
The ones that show up in nearly every suite I see:
None of that means the tool is bad. It means the tool reports what it can see, and whether a finding is a defect depends on intent, which is not in the DOM.
So the failure mode of gating on everything is not false confidence, it is abandonment: the gate fires on something nobody agrees is a bug, somebody adds an exception, then a wildcard, then a skip, and within two sprints you have neither the gate nor the report.
What survives contact with a team is smaller: pick a short list of rules your team has actually agreed are never acceptable, fail on those, and attach the rest so the trend is visible. Grow the list when a category stops appearing.
Wrap commands and output in triple backticks and they will render as a code block.
Checking whether you’re signed in…
Have a different question? Ask it.