agentleFS
jscott3201jscott3201/yet-more-skills/skills/ui-design-review/SKILL.md

ui-design-review

Skill4 starsChanged 2 months agoApache-2.0View on GitHub

Design, polish, or critique a product UI with deliberate hierarchy, typography, layout, states, and responsive behavior. Explicit design pass; preserve established product tokens unless change is requested.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the brief, existing screen, shared tokens, neighboring components, and real content shape.

    line 10
  2. Distinctive:

    Use spacing and type to establish hierarchy.

    line 26
  3. Distinctive:

    Read practice notes and sources only when the task needs more depth.

    line 50

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

UI Design and Visual Review

Ground the screen in its job

Read the brief, existing screen, shared tokens, neighboring components, and real content shape. Determine what the operator needs to notice, understand, and do next. For an existing product, visual consistency and clarity usually matter more than inventing a new identity. Do not turn a workbench into a landing-page hero because a design skill prioritizes novelty.

For genuinely new visual direction, make a compact choice about hierarchy, density, typography, and palette, then implement it. Do not create a separate design system document or mandatory planning artifact unless requested. Respect explicit user aesthetics over generic preferences.

Design all meaningful states

Work with realistic long names, counts, timestamps, units, and permission states. Include loading, empty, error, unavailable, stale, selected, disabled, and pending outcomes as applicable. Use concise action labels and specific recoverable errors. Do not invent business metrics or customer evidence to make a mock screen look populated; label illustrative data clearly.

Use spacing and type to establish hierarchy. Reserve emphasis for the primary task and meaningful status. Avoid decorating every datum as a card, hiding important controls in tooltips, or using color alone for operational meaning. Motion should explain a transition, respect reduced motion, and not compete with live data.

Walk through a real task

Choose a representative task with a clear start and observable finish. Check whether the user can find the next action, understand the consequence, recognize completion, and recover from one realistic error without losing useful work. Repeated backtracking, hidden prerequisites, and unclear outcomes matter more than cosmetic novelty.

Keep inspection findings separate from usability evidence. A source or browser walkthrough can suggest a problem; it does not prove a measured improvement in task success or speed. Use realistic localized content and narrow layouts when those are supported, rather than evaluating only ideal short English labels.

Implement, inspect, refine

Reuse framework-specific shadcn/local components. Inspect the rendered result when browser tools are available: desktop and narrow view, both supported themes, long content, open overlays, and failure states. Check alignment, overflow, contrast, focus visibility, and text hierarchy. A screenshot alone does not prove interaction behavior.

Make a focused refinement pass and explain only consequential design decisions. Without a rendered browser, report a source-level design review and leave visual validation explicitly unrun. Do not claim pixel-perfect output or accessibility conformance from code inspection.

Further guidance

Read practice notes and sources only when the task needs more depth.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.