CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

common-ui-design

Design distinctive, production-grade frontend UI with bold aesthetic choices. Use when building web components, pages, interfaces, dashboards, or applications in any framework (React, Next.js, Angular, Vue, HTML/CSS).

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

The canonical home for this skill is common-ui-design in HoangNguyen0403/agent-skills-standard

SKILL.md
Quality
Evals
Security

UI Design Direction

Priority: P0 (CRITICAL)

Commit to deliberate aesthetic direction grounded in task, audience, and brand before writing code.

Phase 0: Design Thinking & Journey Hierarchy (Pre-Code)

Answer these before choosing components or layouts:

  1. Context & Baseline: For existing designs, polish is the default (preserve structure/behavior). Restyle and rethink BOTH require explicit human request and approval; restyle preserves flow and content behavior, rethink alters structure. For new designs, explore purposefully without an absent baseline constraint.
  2. Audience & Primary Journey: Establish primary user journey, required decisions, content order, and dominant action before picking components. Reject automatic hero → equal-card → pricing templates without task purpose. Cards group genuinely independent items; never enforce generic symmetry over hierarchy.
  3. Character & Tone: Define 2–3 concrete traits grounded in brand and domain (e.g., utilitarian, editorial, refined, playful). Intentionality over extreme novelty. Restrained neutral design is valid when utility demands it.
  4. Visual Purpose & Restraint: Visual hierarchy must direct attention to task completion; preserve required helper text, safety notices, and keyboard focus destinations.

Aesthetic & Design System Foundations

Typography & Spatial Composition

  • Match typography to personality and platform conventions. Platform fonts (system-ui, SF Pro, Roboto) are valid for high-density tools; pair distinctive display + refined body fonts when identity demands character.
  • Ensure readable type scale, robust hierarchy, and accessible line lengths.
  • Self-host fonts via framework loaders or @font-face — never use CDN <link> in production.
  • Structure density by purpose: generous negative space for focus; compact rhythm for data. Grid breaks must serve flow, not ornament.
  • See Tone Palette & Font Pairings

Role-Based Color, Tokens & Contrast

  • System First: Use existing project design systems first. If absent, define minimal role-based tokens: brand accent vs semantic roles (success, warning, error, info); canvas, surface, overlay, border, focus; paired text/icon foregrounds; control radii and elevation.
  • Harmony & Relationships: Evaluate undertones, saturation, lightness, and meaningful surface separation with visual emphasis proportional to task importance. No fixed palette ratios, color counts, or mandatory OKLCH.
  • Composited Contrast: WCAG AA on actual browser-composited pixels for solid fills, alpha overlays, and gradients across all intermediate glyph regions and state cues (normal text ≥ 4.5:1, large text ≥ 3:1, non-text cues ≥ 3:1).
  • Theme & States: Dark themes require independent surface/text mappings, not color inversion; respect forced colors. Non-color-only state cues (icons/text/patterns). Cover interactive states (hover, focus, pressed, disabled, loading).

Motion

  • One well-orchestrated transition > scattered micro-interactions. CSS-first; always honor prefers-reduced-motion. See Motion Patterns.

Rendered Critique & Human Refinement

  • Rendered Inspection: Critique actual screenshots and populated viewports with state and region specificity.
  • Loop: Observation → Impact → Smallest Change → Preserve Constraint. Distinguish verifiable facts from impact hypotheses.
  • Classification: Defects (must fix), context concerns (recommend), taste (human judgment decides; original or preference can win without objective proof).
  • Three Separate Delivery Receipts: Separate functional smoke (interaction/keyboard/contract), human visual acceptance (maintainer appearance sign-off), and model efficacy (controlled empirical benchmark).
  • See Human Direction & Refinement Protocol

Anti-Patterns

  • Reject unjustified template defaults: Do not default to hero → three equal cards → pricing grid when unsupported by task.
  • No unreasoned style bans: Do not ban colors (e.g. brand-appropriate purple), cards, or platform fonts; ground choices in task and brand.
  • No unapproved restyle/rethink: Never dismantle existing layouts or recovery flows without explicit human direction.
  • No uncomposited contrast shortcuts: Never test gradient/alpha contrast at endpoints alone or convey status by color alone.
  • No false equivalence: Never treat passing automated tests as human visual acceptance or model efficacy.

References

Repository

HoangNguyen0403/agent-skills-standard

Path

.agents/skills/common/common-ui-design/SKILL.md

Last updated
First committed

Canonical home

HoangNguyen0403/agent-skills-standard
In sync

since Apr 14, 2026

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.