Responsive and accessibility evidence

This page separates automated browser evidence from device and assistive-technology checks. A passing desktop emulator is useful evidence, but it is not a substitute for a real device or screen-reader session.

Automated browser evidence

The release gate runs type checking, a static build, content and route checks, Lab unit tests, the full Chromium Playwright suite, and focused responsive and keyboard suites in Playwright WebKit and Firefox. The browser suite checks the homepage, tutorials, validation, and Lab at 320, 375, 390, and 430 CSS pixels, plus the compact header controls at 500 pixels.

Deployed verification record

The production deployment is created only after the CI workflow succeeds for the same revision.

Deployed revision: 7a23b56acd64 · Open CI run

What the automated suite checks

Layout

No page-level horizontal overflow at the tested viewport widths.

Mobile information density

At narrow widths, the working molecule depiction remains first and calculated properties use an explicit expandable control.

Keyboard

Mobile menu Enter, Space, and Escape handling; the mini-demo properties summary with Enter and Space; notebook arrow-key tabs; and dialog focus restoration.

Screen-reader semantics

Axe checks for the expanded menu, dialogs, Lab results, and light/dark pages. Playwright also checks concise Lab status updates for view changes, investigation choices, safety gear, and results.

Motion

Reduced-motion behavior keeps Lab results understandable without animation.

Browser engines

Chromium runs the full browser suite. Playwright WebKit and Firefox run focused responsive and keyboard checks; neither is device evidence.

Inspect the published verification protocol

Read the browser checks and manual-device protocol published with this site. These links describe the checks; they do not turn unrecorded device evidence into a passing result.

What this does not prove

Chromium automation does not prove touch ergonomics in iPhone Safari or Android Chrome, Safari keyboard behavior, Windows high-contrast mode, or NVDA announcements. Those checks remain manual release evidence.

Manual device and assistive-technology record

Two scoped macOS Safari observations are published below. A complete device or assistive-technology record has not been published yet.

Recorded partial observation

22 September 2026

On the production HTTPS page in desktop macOS Safari, the native accessibility tree exposed the skip link, primary navigation, language and theme controls, the SMILES input label, and analysis output labels.

What this does not prove: Not yet recorded: comprehensive Safari keyboard flows, viewport resizing, touch interaction, Android Chrome, and screen-reader announcements. This observation is not a device-support claim.

Recorded Safari keyboard observation

23 September 2026

On the production HTTPS homepage in desktop macOS Safari, keyboard activation with Enter followed the skip link to #content and moved focus to the main content area.

What this does not prove: This verifies only the skip-link path on desktop Safari. The compact menu, tabs, dialogs, viewport resizing, touch interaction, Android Chrome, and screen-reader announcements remain unrecorded. This is not a device-support claim.

EnvironmentViewportChecks
iPhone Safari375/390 px, portrait and landscapeTouch targets, header, mini-demo properties, code blocks, tutorial details, Lab next action
Android Chrome360/412 px with text zoomHeader, dialogs, horizontal scrolling
macOS SafariDesktop keyboard onlySkip link, narrow-window menu, tabs, dialogs
Windows Chrome + NVDADesktopMenu labels, tab changes, dialog labels, Lab result announcements

Manual checks required before a device claim

  • 200% zoom reflows without losing controls or page-level horizontal scrolling.
  • Tab, Shift+Tab, Enter, Space, and Escape complete the menu and dialog flows.
  • Arrow keys, Home, and End select the expected Lab notebook tab.
  • Reduced-motion and high-contrast modes keep focus, controls, and result text distinguishable.

Read chemistry validation and benchmark evidence