What problem does it solve? Turning a vague UI idea into a concrete, agreed visual direction is slow when it happens only in prose. This Skill renders the user-visible surface of a feature as static HTML mockup pages, publishes them as reviewable Claude Artifacts, and iterates with the user until one direction is accepted, returning the decisions to the calling design skill. ## Core Features & Use Cases - Grounded mockup generation: Reads the host project's existing design language (components, tokens, prior mockups) and produces self-contained, offline HTML pages that look like they belong in the project. - New vs. modify workflows: A brand-new surface gets one overall view; a modification to an existing surface gets 2-4 uniquely named alternatives that differ on a real axis such as placement, density, or typography. - Artifact publishing and iteration: Each page is published as a Claude Artifact with a stable URL, revised in place across feedback rounds, and the accepted result is returned as a parseable decision block to the calling skill. - Use Case: While designing a reminders feature for a todo app, the design skill invokes this Skill to render the reminder list and editor as an HTML mockup, the user picks a right-hand drawer layout, and the concrete UI decisions flow directly into the design document. ## Quick Start This Skill is not user-invocable; it runs automatically when the feature-design skill reaches its mockup step for a feature with a user-visible surface.
