The same mount source, editable, with Radix UI resolved at the exact scored version. Example styling ships as polish.css; delete its import for the bare mount. Runs on CodeSandbox's bundler; the measurement belongs to the untouched mount.
import*asTabsfrom"@radix-ui/react-tabs";import{TEXT}from"./harness-kit";/**
* Radix UI tabs, mounted exactly as the Radix documentation describes.
*
* Panels contain text and nothing focusable, per the protocol, so the APG
* requirement that an otherwise-empty panel be reachable is actually testable.
*/exportfunctionTabsHarness(){return(<><buttondata-testid="hr-before"type="button">
Before
</button><Tabs.RootdefaultValue="one"><Tabs.Listdata-testid="hr-tablist"><Tabs.Triggervalue="one"data-testid="hr-tab-1">{TEXT.tab1}</Tabs.Trigger><Tabs.Triggervalue="two"data-testid="hr-tab-2">{TEXT.tab2}</Tabs.Trigger><Tabs.Triggervalue="three"data-testid="hr-tab-3">{TEXT.tab3}</Tabs.Trigger></Tabs.List><Tabs.Contentvalue="one"data-testid="hr-panel-1">{TEXT.panel1}</Tabs.Content><Tabs.Contentvalue="two"data-testid="hr-panel-2">{TEXT.panel2}</Tabs.Content><Tabs.Contentvalue="three"data-testid="hr-panel-3">{TEXT.panel3}</Tabs.Content></Tabs.Root><buttondata-testid="hr-after"type="button">
After
</button></>);}
Self-run companion check, approximated in-browser from ARIA attributes and visibility. Official scores are measured by the runner with replayable Playwright traces.