Lantern

What it does

Lantern holds a light up to your work and shows what’s hidden in the dark: the things that stop disabled people using it. It checks against WCAG 2.2 Level AA and writes up each issue by severity, with a fix you can make today.

How it works

1. Ask Claude. Say “is this accessible?” with a URL, your local dev server, a Figma link, a screenshot or some code.
2. Lantern looks closely. On a page, it runs axe-core, then does what automated tools miss: a keyboard-only pass, visible focus, traps, 320px reflow, 200% text zoom, reduced motion, target sizes and focus hidden behind sticky bars. In Figma, it works out contrast from the real fills, gradients included, and checks every component has a focus state.
3. You get a report. Blockers first, each with the WCAG criterion, where it is, who it affects and the fix. “Passed” and “Not checked” lists mean silence never looks like a pass.
4. Take it to your tickets. Ask for a CSV and paste the issues straight into your tracker.

What it checks

• Live pages and dev servers: axe-core plus a manual pass over keyboard, focus, layout, motion, targets, names, labels and errors.
• Figma frames: contrast from the actual fills, target sizes, focus and error states, colour-only meaning, and reading order.
• Screenshots: what’s visible, clearly marked as estimated, plus an honest list of everything a picture can’t show.
• Code: HTML, JSX, Astro and Vue, checked for semantics and ARIA mistakes, with file and line for each.
• Quick questions: “Does this grey pass on white?” gets a ratio, a yes or no, and a nearby colour that passes.

Lantern never invents a finding. If it can’t be sure, it tells you how to check.