← ClaudeAtlas

art-directionlisted

Design an interface with a visual language of its own instead of assembling one from defaults. Use when building a landing page, product UI, portfolio or marketing site where the look has to be distinctive — and whenever a design risks reading as template-generated. Derives the visual language from the subject, writes ART-DIRECTION.md and DESIGN-TOKENS.md before any code, then audits the built result against them. For products with flows or forms, run ux-direction first.
audient-id4/design-skills · ★ 0 · Web & Frontend · score 67
Install: claude install-skill audient-id4/design-skills
# Art direction Most generated interfaces fail the same way. Not ugly — *anonymous*. Inter, `rounded-xl`, a soft shadow, three equal cards, a violet gradient, a hero with a subtitle and a button. Every decision defensible, the whole thing forgettable. The cause is not bad taste. It is that no decision was ever *made*. Defaults filled every slot before anyone asked what the thing should feel like. This skill forces the question first, records the answer, and then checks the built result against it. ## Dependency This skill defines **visual language**. It does not decide user flows, information architecture or interaction priorities. If `UX-DIRECTION.md` exists, it is upstream — read it first and treat it as settled. The primary action it names is what the composition must make dominant; the hierarchy it defines is the hierarchy the design expresses. If it does not exist, make reasonable assumptions about behaviour and write them down in `ART-DIRECTION.md` under **Assumed behaviour**, so the next person can see what the visual system was designed against and correct it. A landing page usually needs only this skill. Anything with flows, forms, empty states or destructive actions wants `ux-direction` run first. ## Functionality wins Read this before the rest, because everything below is subordinate to it. Experimental design is permitted only where it improves understanding, emotion, usability or brand memory. Novelty that costs any of those is not a trade-off worth ma