Prism is an experimental Claude Code skill exploring how AI can support product designers during design critique, accessibility review, and handoff preparation. The idea came from a simple observation: interface quality is rarely one-dimensional. A screen can look visually polished and still fail accessibility. A flow can technically satisfy a checklist and still feel confusing. A component can be implemented correctly and still create friction in the actual user journey.

I built Prism to bring these perspectives closer together. Instead of treating UX/UI review and accessibility inspection as two separate activities, Prism combines them into one structured workflow. It helps evaluate an interface through design quality, usability risks, accessibility signals, and handoff readiness. The skill can review screenshots, Figma links, live sites, or HTML files, and depending on the context, it can run a UX/UI inspection, a WCAG 2.2 AA accessibility review, or a combined interface quality review.

The UX/UI inspection focuses on visual hierarchy, cognitive load, interaction patterns, consistency, usability risks, and established design principles such as Fitts’s Law, Hick’s Law, Miller’s Law, Nielsen’s heuristics, Norman’s emotional design model, Gestalt principles, and others. The accessibility inspection focuses on contrast, keyboard navigation, focus visibility, semantic structure, heading hierarchy, ARIA usage, screen reader risks, and WCAG 2.2 AA criteria. For live sites and HTML files, Prism can inspect actual markup and CSS instead of relying only on visual assumptions from a screenshot.

The goal is not to replace design judgment. The goal is to make interface review more explicit, more consistent, and easier to repeat across product teams — especially before handoff, during design review, or when validating product screens against both usability and accessibility expectations.

Prism is part of my broader work around design systems, design-to-code workflows, and AI-assisted product quality checks. It reflects a direction I care deeply about: moving design work beyond static mockups and toward more structured, testable, and production-aware ways of evaluating digital products.