FAQ
Questions, answered plainly.
If something here is still unclear, ask us directly — we would rather be precise than persuasive.
- How is this different from an accessibility scanner?
- Rule checkers inspect the DOM against a fixed rule set. They cannot tell you that your select component stopped announcing its options. Coverli runs the real screen reader, captures what is spoken, and diffs it against an approved baseline. We run deterministic checks too — they just aren't the whole story.
- Do you really run NVDA and VoiceOver, or is it a simulation?
- Real NVDA on real Windows machines and real VoiceOver on real Macs, provisioned per run. We capture text at the speech-synthesizer layer, so there is no audio transcription and no approximation.
- Won't the diffs be flaky?
- Announcements are normalized before comparison — volatile IDs, timestamps, and synthesizer chatter are stripped. A baseline only changes when the spoken experience changes.
- Will this slow down our pipeline?
- A fast semantic tier covers every commit in minutes and the real-AT tiers act as ground truth. Change detection fingerprints each component build, so unchanged stories are skipped entirely.
- What do we need in place to start?
- A Storybook. Our GitHub App builds it, walks every story, and reports back as a pull-request status check.
- Can we use the report for compliance?
- Yes — as evidence. Reports cite WCAG 2.2, quote your source at file-and-line, and are tied to a specific commit. Coverli produces conformance evidence, not a conformance claim; the legal determination stays human.
- What does a partial audit mean?
- If only part of your system was in scope, the report says so explicitly and the ship-status verdict is scoped to that coverage. We never imply we looked at something we didn't.
- Does Coverli change our product at runtime?
- Never. We are not an overlay. We test your source and report defects so you can fix them at the source.