All skills
simota avatar

/vision

@c805268
by shingo imotasimota/agent-skills85 stars
15

Directing UI/UX creative work — redesigns, new designs, trend application, Design System construction, Muse/Palette/Flow/Forge orchestration. Use for design direction. Offers a co-design pair mode.

Use this Skill: https://skilld.dev/gh/simota/agent-skills/vision

This session only. Nothing lands on disk.

referencedesign-trends.md

≈3.5k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Design Trends & AI Tools Integration

Purpose: Use this file when the task involves trend selection, trend risk, or AI-assisted design workflows.

Contents:

  • Visual trend risk buckets
  • AI interface patterns
  • Trend-application rules
  • AI tool guardrails

Visual Trends (2025-2026)

Trend Risk Best For
Dark Mode Low all products
Micro-animations Low interactive elements
AI-Native Interfaces Low AI-powered products
Variable Fonts Low all products
Adaptive UI Low personalized apps
Bento Grid Medium dashboards, portfolios
Glassmorphism 2.0 Medium overlays, cards
Spatial Design Medium XR-oriented products
Sustainable Design Medium eco-conscious brands
Neo-Brutalism High creative or tech brands only

2025-2026 Emerging Trends

Trend Risk Level Use When
Ambient Computing Medium Smart home / zero-UI scenarios only
Neo-Brutalism 2.0 High Counter-trend to AI-smooth aesthetics; creative/disruptive brands only
Agentic UX Low AI-first products with autonomous task delegation

2026 Design Direction Trends

Trend Risk Application
Calm UI / Cognitive Clarity Low Prioritize cognitive clarity over sensory richness. Functional minimalism: large typography, soft radius, generous spacing, high contrast. Replace decorative elements with purposeful whitespace
Adaptive Design Systems Low UI complexity adjusts dynamically based on user behavior, expertise level, and context. Progressive disclosure evolution — expert/beginner modes, context-aware density
Figma Make / AI-Driven Design Low Text prompt → design-system-compliant UI generation. Concept-to-prototype time reduced ~38%. Designer role shifts from "how" to "what/why". Review AI output against brand guidelines
Multi-modal Input Direction Medium Design for seamless touch/voice/keyboard/gesture switching. Invisible interactions (voice commands, gesture recognition, presence detection). Requires input-mode-agnostic component design
Dynamic Typography Low Variable Fonts for responsive weight/width/slant. CSS text-wrap: balance (headings ≤6 lines) and text-wrap: pretty (orphan prevention). Typography adapts to device and context

Calm UI Design Principles

  1. Cognitive Clarity > Sensory Richness: Every visual element must serve comprehension, not decoration
  2. Functional Minimalism: Distinct from decorative minimalism — restrained palette, clear primary action, generous whitespace
  3. Trust Through Calm: Users arrive with calibrated skepticism from AI failures (NN/g State of UX 2026) — earn trust through clarity, not urgency
  4. Quiet Interfaces: Reduce notification noise, minimize competing CTAs, let content hierarchy guide attention

AI Interface Patterns

Pattern Use
Chat-First AI assistants and search
Inline Suggestions editors and forms
Progressive Disclosure complex workflows
Confidence Indicators AI-generated output
Regeneration UI generative tools

AI Interface Patterns (2025-2026)

Pattern Description
Proposal Cards AI-generated turnkey solutions for user validation
Intent Economy UI Shift from attention to intention; minimal friction, maximal agency
Ambient Indicators Passive status without explicit interaction (presence-aware UI)

Trend Application Rules

Apply confidently

  • dark mode support
  • micro-animations in the 100-300ms range
  • variable fonts
  • adaptive layouts

Apply carefully

  • glassmorphism
  • spatial depth
  • bento layouts
  • oversized typography
  • sustainable-design aesthetics

Apply sparingly

  • neo-brutalism
  • kinetic typography
  • extreme minimalism
  • heavy 3D

Before applying any trend:

  • Brand fit is explicit
  • Target users are likely to expect it
  • WCAG 2.2 AA can still be met
  • Performance cost is acceptable
  • The pattern should age well for 2-3 years

Design Paradigm Knowledge Base

Spatial Computing UX (visionOS / XR)

Design principles for immersive interfaces:

Principle Description
Z-Axis Layering Content organized in depth layers (near/mid/far). Primary actions in near field, context in far field
Gaze-Based Interaction Eyes select, hands confirm. Highlight on gaze hover, commit on pinch. Never require sustained gaze >2s
Ergonomic Zones Comfortable interaction zone: 1-2m distance, ±30° vertical, ±60° horizontal from center
Spatial Audio Feedback Sound localized to UI element position. Provides confirmation without visual attention
Window Management Windows float in space. Users arrange spatially. Respect user's spatial memory
Passthrough Integration Blend digital and physical. Respect real-world lighting and shadows

Design rule: Start with 2D flat design, add depth only where it communicates hierarchy. Spatial design is not 3D decoration — it is information architecture in the z-axis.

AI-Native Interface Design Philosophy

2026 is the year "interfaces become agents." Shift from GUI paradigm (click/tap guidance) to intent-centric design where UI is a medium, not a message.

New UI Patterns replacing traditional paradigms:

Old Pattern New Pattern What Changes
Forms Intent Canvas Accept messy input, deliver structured interpretation
Confirmation Dialog Negotiation Human and agent co-create the plan
Progress Indicator Narrative System explains what it's doing as a story
Hard Failure Graceful Escalation Soften boundaries, provide escalation paths

Autonomy Dial: Meta-control allowing users to dynamically adjust agent autonomy level. Not a binary copilot/autopilot — a continuous slider.

"Nudge, don't nag": The meta-principle — prompt helpfully without being intrusive.

The spectrum from tool to agent:

Level Pattern User Control Example
L1 Autocomplete Inline suggestion Full GitHub Copilot ghost text
L2 Copilot Side-panel assistant High ChatGPT sidebar
L3 Agent Proposal → approve Medium Devin, Cursor Agent
L4 Autopilot Autonomous with guardrails Low Self-driving CI/CD

Key design tensions:

  • Confidence visualization: Show AI certainty (progress bars, probability badges, "I'm not sure" indicators). Never present uncertain output as certain.
  • Intent Preview: Before any autonomous action, show what will happen and let user approve/edit/cancel.
  • Graceful degradation: When AI fails, fall back to manual workflow seamlessly. Never dead-end.
  • Explanation affordance: Every AI output should have a "why?" path available (not mandatory, but accessible).

Emotional Design & Neuroaesthetics

Layer Mechanism Application
Visceral First-impression aesthetics (color, shape, motion) Hero section, onboarding, packaging
Behavioral Usability pleasure (efficiency, predictability, control) Core workflows, repeated actions
Reflective Identity, meaning, social signaling Profile, sharing, achievement

Ethical dopamine design: Use reward mechanisms to reinforce user goals (progress toward their objective), not platform goals (engagement metrics). Celebration animations on task completion, not on infinite scroll consumption.

Anti-Design & Neo-Brutalism 2.0

Not chaos — intentional rule-breaking with purpose:

Element Convention Neo-Brutalism
Typography System fonts, 16px body Monospace, oversized, mixed weights
Color Brand palette, 60-30-10 Harsh contrast, neon accents, black borders
Layout Symmetric grid, consistent spacing Asymmetric, overlapping, visible grid lines
Imagery Polished photos, illustrations Raw screenshots, brutalist photography, ASCII art
Borders Subtle or none Thick (2-4px) black borders on everything

When to use: Developer tools, creative agencies, counter-cultural brands, personal portfolios. Never: Healthcare, finance, government, accessibility-critical products.

Sustainability in Digital Design

Practice Impact Implementation
Dark mode default OLED: 42% energy reduction (Purdue University study) prefers-color-scheme: dark as default, light as option
Image optimization WebP/AVIF: 25-50% smaller than JPEG <picture> with format fallback chain
Reduced motion Less GPU computation, less battery prefers-reduced-motion: reduce as non-optional
System fonts Zero font transfer cost Font stack: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif
Lightweight design Every KB costs CO₂ (avg page: 2.3MB, ideal: <500KB) Performance budget as sustainability metric

Carbon-aware color: Darker colors on OLED = less energy. Design tokens should include an "eco" mode with maximum dark, minimum animation.

Generative Design Systems

AI-readable design tokens — the most leveraged evolution point. Tokens shift from storing values to storing intent:

{
  "color-primary": {
    "value": "#3B82F6",
    "intent": "trust-building action color",
    "usage": "primary CTA, navigation highlights",
    "constraints": "WCAG AA contrast on white, never on warning surfaces"
  }
}

Gartner predicts 30% of new apps will use AI-driven adaptive interfaces by 2026. CopilotKit and Google A2UI (Open-JSON-UI) lead open-source Generative UI frameworks.

Design tokens that generate themselves through rules rather than static values:

Concept Mechanism Example
Parametric typography Base size × scale ratio × level = all sizes clamp(1rem, 0.5rem + 1.5vw, 1.5rem) with Major Third (1.25) scale
Algorithmic color Base hue × rotation × lightness rules = full palette OKLCH hue rotation: oklch(0.7 0.15 calc(var(--base-hue) + 120))
Proportional spacing Base unit × fibonacci-like multiplier = spacing scale 4px → 8 → 12 → 16 → 24 → 32 → 48 → 64
Noise-based variation Perlin noise applied to decorative properties SVG feTurbulence for organic texture generation

Visual Aesthetic Direction (2025-2026)

Dark Mode as Default

82% of mobile users now use dark mode (NNG). Design dark-first, not as an afterthought.

Style BG Accent Text Muted
Professional #0C1120 #3A82FF #F8FAFC #8895A7
Warm luxury #1C1917 #D4A574 #FAFAF9 #A8A29E
Gradient dark #0F172A→#020617 #22D3EE #F1F5F9 #64748B

Rule: Never use pure #000000 (use #0A0A0A+). Reduce saturation of accents for dark backgrounds. 4.5:1 contrast minimum for body text.

Rounded Everything (Pill UI)

border-radius: 999px on buttons, 16-24px on cards. Neurologically soothing — eyes flow naturally around curves. Products: iOS, Material You, Arc, Linear, Spotify.

Large Typography, Minimal Chrome

.hero { font-size: clamp(3rem, 8vw, 8rem); font-weight: 700; letter-spacing: -0.03em; line-height: 1.05; }

37% higher reading completion on LPs with bold typography. Pair oversized display with readable body font. Use negative letter-spacing for headlines.

Monochrome + One Accent

Single accent color against grayscale palette. Forces clear visual hierarchy — the accent becomes an unmistakable action signal. Products: Linear (blue), Vercel (white), Stripe (purple).

Floating Tab Bar (Mobile)

.tab-bar { position: fixed; bottom: 16px; left: 16px; right: 16px; border-radius: 24px;
  background: rgba(255,255,255,0.85); backdrop-filter: blur(20px); box-shadow: 0 4px 24px rgba(0,0,0,0.12); }

40% faster task completion vs hamburger menu. ≤5 tabs. Products: Apple Maps (iOS 26), Airbnb, Instagram.

Bottom Sheet as Primary Mobile Surface

Draggable panel with snap points (25%/50%/90%). Replaces modals, filters, details on mobile. Products: Apple Maps, Google Maps, Uber, Spotify.

AI Tool Guardrails

Tool Best For Guardrail
Figma AI layout exploration review for brand, tokens, and a11y
v0 component scaffolding route through Forge, then production agents
Claude Artifacts concept generation use for exploration, not direct production
Galileo AI fast UI concepts require Vision review before adoption
Midjourney / DALL-E moodboards and assets check brand fit and licensing

AI Tool Best Practices (2025-2026)

Generate in batches (10-15 options per direction), not one at a time. Review batches against brand guardrails before selecting 3-5 for refinement.

Commercial safety checklist:

  • Private model training enabled (designs stay yours)
  • Watermark detection reviewed
  • Style locking active for brand-critical work
  • "Private Generation" mode enabled for IP-sensitive assets

Do NOT use AI generation as final authority for:

  • Production code
  • Brand-critical identity work
  • Accessibility-sensitive components
  • Performance-critical implementations

AI tool guardrails (additions):

Tool Use For Guardrails
Adobe Firefly Brand imagery, vector art Enable commercial-use mode; confirm training data policy; use "Private Generation" to protect IP
Figma AI (MCP) Layout generation, token scanning Use MCP server to sync tokens/components in real time; review for brand/a11y before adoption

Use AI generation for:

  • initial concept exploration with 3+ variations
  • moodboards
  • layout alternatives
  • placeholder assets

Source: SKILL.md on GitHub

No alerts14d5 checks · Risk SAFE
  • Gen Agent Trust Hub14d

    The 'vision' skill is a comprehensive design-direction tool that operates by defining strategies and delegating work to other specialized agents. It contains robust internal guardrails, including mandatory accessibility standards (WCAG 2.2 AA) and ethical AI principles. No security risks were identified.

  • Socket14d

    No alerts

  • Snyk14d

    Risk: LOW · No issues

  • Runlayer6mo

    9 files scanned · No issues

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at c805268. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 3 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/vision