Frontend design
Create distinctive, production-grade frontend interfaces with strong visual direction, polished typography, considered layout, and working HTML/CSS/JS or framework code. Use for websites, landing pages, dashboards, React components, application screens, and UI beautification.From its SKILL.md
npx -y skills add nexu-io/open-design --skill frontend-designAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its file declares
Copied from the file, not written here
The file declares its own license as Complete terms in LICENSE.txt. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
5.1 KB, 867 tokens by cl100k_base, as published. Nobody here has run it
frontend-design
Adapted from Anthropic's official
frontend-designskill for Open Design.
Use this skill when the user asks to build or improve a frontend interface: a website, landing page, dashboard, application screen, HTML/CSS artifact, React/Vue/Svelte component, or a visual redesign of an existing UI.
The goal is not just "make it nicer." The goal is to ship working frontend code with a clear design point of view, strong craft, and enough product detail that the result feels designed for the user's actual context.
Workflow
-
Understand the brief before choosing the look.
- Identify the audience, primary job, domain, and emotional tone.
- Note any technical constraints: framework, existing styles, accessibility, performance, export target, or responsive requirements.
- If the repo already has design tokens, components, screenshots, or a
DESIGN.md, use those as the source of truth.
-
Commit to one specific aesthetic direction.
- Pick a direction that fits the product: brutally minimal, editorial, luxury, playful, industrial, retro-futuristic, dense operational, calm enterprise, artful consumer, or another precise direction.
- Make the direction concrete through typography, spacing, color, hierarchy, motion, and component shape.
- Avoid generic AI defaults: purple-blue gradients, vague glass cards, interchangeable SaaS layouts, over-rounded cards, stock icon rows, and decorative blobs that do not serve the interface.
-
Design the real interface, not a placeholder poster.
- Include the controls, empty/loading/error states, tables, filters, navigation, and responsive behavior a real user would expect.
- Use honest content. If data is unknown, label it as sample, pending, or unavailable instead of inventing claims.
- Keep workflows efficient for the target user. Dashboards and tools should be scannable and dense enough for repeated use; marketing pages can be more expressive.
-
Build production-grade frontend code.
- Prefer the repository's existing framework, component conventions, icons, tokens, and styling approach.
- For standalone artifacts, create self-contained HTML/CSS/JS unless the user asked for a framework.
- Use semantic markup, keyboard-accessible controls, visible focus states, sensible contrast, and responsive layout constraints.
- Use CSS variables for repeated colors, spacing, shadows, and type scale.
-
Refine visual craft.
- Typography: choose expressive but readable type pairings. Avoid using default system stacks as the main visual idea unless the direction is intentionally utilitarian.
- Color: create a balanced palette with role clarity. Use accent color sparingly and deliberately.
- Layout: use alignment, rhythm, density, and negative space intentionally. Do not let cards, panels, or labels drift.
- Motion: add purposeful transitions for state changes, reveals, and feedback. Prefer transforms and opacity for performance.
- Details: use texture, borders, shadows, dividers, media, and iconography only when they support the concept.
-
Self-review before final delivery.
- The interface works at mobile and desktop widths.
- Text fits its containers and does not overlap adjacent UI.
- Interactive elements have hover/focus/active/disabled states.
- The design avoids obvious AI-generated visual tropes.
- The result has one memorable quality a user could describe after closing the page.
Open Design Integration
When Open Design provides an active design system, treat it as the product's brand contract. Use the injected color, typography, layout, and component guidance first, then apply this skill's frontend craft rules where the design system is silent.
When Open Design injects craft references such as typography, color, and anti-ai-slop, apply those checks before finishing. If the user's brand guidance conflicts with a generic craft rule, the user's brand guidance wins.
Source
- Upstream: https://github.com/anthropics/skills/tree/main/skills/frontend-design
- Category:
web-artifacts
License
This skill is adapted from Anthropic's official skills repository. See LICENSE.txt in this folder for the upstream Apache-2.0 license terms.
What ships with it: 1 file
9.9 KB alongside SKILL.md
- LICENSE.txt9.9 KB
Gives 2 of the 12 instructions most design frontend skills give in 867 tokens
Counted across 1,179 of the 2,086 authors here whose files we hold, read 2026-09-06
- Commit to a bold aesthetic directionhere, and in 31 of 1179, across 24 files
- Prefer component composition over inheritancein 28 of 1179, across 14 files
- Animate only transform and opacity propertiesin 27 of 1179, across 22 files
- Memoize expensive computations with useMemoin 26 of 1179, across 13 files
- Use semantic HTML elementsin 24 of 1179, across 23 files
- Virtualize long lists for performancein 21 of 1179, across 10 files
- Use CSS variables for design tokenshere, and in 20 of 1179, across 14 files
- Implement loading, empty, and error statesin 20 of 1179
- Lazy load heavy components with Suspensein 19 of 1179, across 8 files
- Respect prefers-reduced-motion media queriesin 18 of 1179, across 10 files
- Prioritize CSS-only animations for HTMLin 18 of 1179, across 16 files
- Use compound components for related UI elementsin 18 of 1179, across 7 files
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.