What problem does it solve? Teams often jump straight into visual design or component implementation without agreeing on an app's structural composition, resulting in GUIs that merely redraw the incumbent layout or promise interactions the domain cannot support. This Skill runs an interview-driven concept session that grounds the design in the real domain model and produces a versioned ASCII blueprint document before any visual work begins. ## Core Features & Use Cases - Domain grounding: Builds a cheap object inventory (objects, verbs, invariants, hero objects) from code, docs, or the interview itself before sketching anything. - Structured interview loop: Batched question rounds covering primary verb, user type, feel appetite, and reference apps, with side-by-side ASCII wireframe previews for paradigm selection. - Three-paradigm divergence: Sketches exactly three structurally distinct concepts from metaphor families (document, spatial, process, instrument, game/bench, conversation), banning the incumbent composition. - Versioned blueprint document: Writes a UX-BLUEPRINT.md with screens, object-to-component contract, structural invariants, honest costs, and rejected directions, refined until explicit acceptance. - Use Case: Before building a new dashboard, run the skill to compare a process-pipeline layout against an instrument-panel layout with real domain objects labeled, then accept a v1 blueprint that constrains all later design work. ## Quick Start Ask the AI to run /gui-blueprint to sketch the structural blueprint for your app's primary GUI through a short interview.
