What problem does it solve? Designing a feature's UI before code exists is hard to structure: ideas stay vague, states get forgotten, and the design never connects to what developers actually build. This Skill turns a product brief and existing screens into a concrete UX proposal that directly seeds the build manifest. ## Core Features & Use Cases - Divergent concept generation: Produces 1-3 UI concepts (layout, flow, component breakdown) using How-Might-We reframing and Crazy-8 style sketching, then converges on one with the user. - State-complete screen design: Defines empty, error, and loading states for every screen, not just the happy path. - Consumer-driven data views: Identifies the data views each screen needs, which seed the read-models' view_shape and the manifest's ui blocks downstream. - Use Case: Given a product-brief.md and existing mockups for a new dashboard feature, produce a ux-proposal.md with screens, components, states, and required data views that the build phase consumes directly. ## Quick Start Ask the agent to design the UI for a feature by reading its product brief and any existing screens, then produce a ux-proposal.md with concepts, components, states, and data views.
