Website interface
The public shell uses semantic headings, one main landmark, labeled navigation, breadcrumbs, a skip link, visible focus styles, responsive layouts, and labeled search controls. Automated site QA checks one H1, language, main landmark, image alternative text, accessible names for controls, horizontal overflow, console errors, and failed requests across forty-eight routes.
Game controls
The fifteen launch games use buttons and numeric inputs that can receive keyboard focus and do not require audio. Automated checks exercise each game at desktop and 390-pixel mobile widths, perform an action and a recovery path, and confirm focusability. Pointer, touch, and keyboard behavior are reviewed where the mechanic supports them.
Known limitations
Automated checks do not prove conformance with every accessibility standard or assistive technology. Some game mechanics rely on visual position, color changes, timed feedback, or compact grids. A successful keyboard path does not establish that every state announcement is ideal for every screen reader. No formal third-party accessibility certification is claimed.
Design choices
Text and controls use strong contrast roles, scalable system fonts, touch-sized buttons, and layouts that avoid horizontal scrolling in the tested viewports. Core games avoid required sound and complex drag gestures. Color is normally paired with text, position, or an accessible label, although each substantive design change requires retesting.
Alternatives and classroom use
Adults and educators should offer another response method when motor, vision, reading, attention, or language needs make a game unsuitable. A game score should not penalize a learner for assistive-technology or input differences. Written explanations and worked examples in the related guides provide a non-game route to the same strategy.
Reporting a barrier
Send accessibility reports to [email protected]. Include the exact URL, device, browser, assistive technology when relevant, the action attempted, expected result, observed result, and whether the issue persists after refresh. Do not include sensitive records.