What problem does it solve? AI-generated interfaces tend to converge on generic, average-looking designs because they rely on training-data patterns instead of real product evidence. This Skill grounds every UI design decision in research from 150,000+ real app screens and 6,000+ user flows, preventing generic "AI slop" output. ## Core Features & Use Cases - Research-First Workflow: A four-phase methodology (Discover, Research, Analyze, Design/Implement) that requires querying real products via Refero MCP tools (search_screens, search_flows, get_screen, get_flow, get_design_guidance) before writing any UI code. - Craft Reference Guides: Built-in guides covering typography, color systems, spacing, motion, icons, copywriting, accessibility, and anti-AI-slop patterns with concrete values and checklists. - Quality Gates: Steal lists, persuasion-layer tables, research summaries, and side-by-side comparison tests that validate designs against top-tier product references. - Use Case: When asked to design a pricing page, the agent first searches real pricing screens from companies like Linear and Stripe, extracts specific tactics (exact copy, spacing values, color choices), then implements a design justified by documented evidence. ## Quick Start Ask the agent to design a new onboarding flow for your app and have it research real product examples with Refero before writing any code.
