AGENT SKILL

color

Load before choosing, extending or checking any color: brand palettes, UI and dark-mode tokens, status colors, chart colors or contrast fixes, even a quick question. Covers palette roles and proportions, 11-13 step OKLCH ramps, semantic tokens, WCAG 2.2 contrast with APCA as an advisory check, dark and high-contrast modes, cultural meaning, Pantone, CMYK and RGB specs, and replacing a generic purple-gradient look with something ownable.

License 许可
MIT
最近更新
2026-11-10
Stars
★ 2
Forks
0

01 前言 Front Matter

name
color
license
MIT
metadata
{ "agent-attribution": "computer=vm; tool=Claude Code; version=2.1.304; timestamp=2026-10-20 02:07:10 UTC", "version-timestamp": "2026-10-20 11:40:00 UTC" }
description
Load before choosing, extending or checking any color: brand palettes, UI and dark-mode tokens, status colors, chart colors or contrast fixes, even a quick question. Covers palette roles and proportions, 11-13 step OKLCH ramps, semantic tokens, WCAG 2.2 contrast with APCA as an advisory check, dark and high-contrast modes, cultural meaning, Pantone, CMYK and RGB specs, and replacing a generic purple-gradient look with something ownable.

02 资源文件 6

03 SKILL.md 正文

Color

Excellent color work is mostly neutrals and restraint: a tuned neutral ramp does 80-90% of the job, one accent marks what matters, status colors are unambiguous, every text pair passes contrast, and the brand hue is chosen against the category rather than by default.

Defaults

Decision Default Change when
Palette structure 1 neutral ramp (13 steps) + 1 accent ramp (11 steps) + 4 status ramps Brand needs a secondary accent (max 2 more) or data visualization needs a categorical set
Proportion 60-30-10 as a starting point: 60% neutral surfaces, 30% secondary surfaces and text, 10% accent Product UI: accent under 10% of the surface; bold campaigns and neo-brutalist work flood with color
Authoring space OKLCH; ship sRGB hex with optional Display P3 values behind @media (color-gamut: p3) Print: author in the target CMYK profile or spot colors
Ramps 11 steps (50-950), lightness linear from L 0.975 to 0.165, chroma peaking at 500; neutrals add 850 and 975 Yellows and oranges need a hue shift so dark steps do not turn olive
Neutrals Tinted toward the accent hue or a warm bias, chroma 0.01-0.02 Pure gray for photography-led and Swiss work; warm neutrals for food, hospitality, luxury
Text contrast 4.5:1 body, 3:1 large text (24px, or 18.66px bold), 7:1 in high-contrast mode Never lower; logos and decorative text are the only exemptions
Non-text contrast 3:1 for input borders, icons, focus rings, chart marks against adjacent colors Decorative dividers may go lower (--color-border-subtle)
APCA (advisory) Body Lc 75+, preferred Lc 90; Lc 60 for large or UI text; Lc 45 for headlines Use it to tune dark mode and thin type; WCAG 2.2 stays the pass/fail baseline
Focus ring 2px+ solid --color-focus-ring, 3:1 against the background and the component High-contrast mode: thicker ring and system colors in forced-colors
Status hues Success about 150, warning about 75, danger about 22-27, info about 255 (OKLCH hue); fg/bg/border from one ramp Brand accent sits on a status hue: shift the status hue 15-25 degrees or change lightness
Dark mode Canvas L 0.15-0.20, about +0.045 L per elevation level, text L 0.92-0.95, accent moved to a light step (300) with dark text on it OLED-first media apps may use a true black canvas
Brand hue Off the category default unless strategy says belong; avoid indigo/violet (hue 270-300) as an unexamined default The brand already owns the hue; the brief rewards category fit
Gradients 2-3 stops, neighboring hues, in oklch interpolation, never behind body text without a solid backing Gradient is the brand's core asset (style.vibrant-gradient-saas)
Print specs Pantone C and U, CMYK per named profile, RGB, hex Digital-only brands may skip Pantone

Workflow

  1. Read locks and context. Existing brand colors are fixed primitives; extend around them. Run python3 ../../library/scripts/atlas.py query "<brief>" for the style, industry and market entries and read their palette, category conventions and color symbolism.
  2. Define roles before hues with palette architecture: canvas, surfaces, text levels, borders, accent states, status, focus, overlay, data.
  3. Choose the brand hue with brand color: map competitors on a hue wheel, pick an ownable position, test it in a real composition.
  4. Generate ramps with OKLCH ramps: python3 ../../library/scripts/color_tools.py ramp --hue <H> --chroma <C> --name <name>. Match the brand color to its nearest step.
  5. Map ramps to tokens. Fill the semantic set (--color-bg-*, --color-fg-*, --color-border-*, --color-accent-*, --color-{success,warning,danger,info}-{fg,bg,border}, --color-focus-ring, --color-overlay-scrim) for light, dark and high-contrast.
  6. Check contrast for every pair that ships, per contrast: python3 ../../library/scripts/color_tools.py contrast <fg> <bg>, then apca for advisory tuning. Fix failures by moving ramp steps, never by suppressing the check.
  7. Build dark mode per dark mode, not by inverting.
  8. Check meaning per meaning and culture for the target markets; make sure nothing relies on color alone.
  9. Render and inspect. Real screens or layouts in light, dark and high-contrast; grayscale screenshot to test value hierarchy; color-vision-deficiency simulation (Chrome DevTools Rendering panel). Use the design-critique skill's render loop.

Decisions that change the answer

  • Style: each atlas style has a palette recipe (Swiss: ink, paper, one red; quiet minimal: warm neutrals and one muted accent; neo-brutalist: saturated flat fills with black outlines; classic luxury: near-black, ivory, a metallic accent used sparingly).
  • Industry: sets category defaults to work with or against (finance blue and green, health blue and teal, food warm hues); see each industry entry's Category conventions and Where to differentiate.
  • Market: color associations and regulated colors vary; read market.* Color and symbolism.
  • Audience: older adults and low-vision users need higher contrast and no low-chroma pastel text; children's products tolerate more saturation; expert users need calm neutrals and strong status colors.
  • Tier: luxury uses fewer, darker or quieter colors and more neutral space; value uses saturated, high-contrast promotional color.
  • Channel: screens are sRGB or P3; print is CMYK or spot with dot gain; signage needs contrast at distance and under ambient light; email needs colors that survive client dark modes.

Common failures

  • Default indigo or purple-to-pink gradient. Fix: choose the hue from the category map in brand color.
  • Pure gray neutrals with a saturated accent. Looks unfinished. Fix: tint neutrals 0.005-0.015 chroma toward the accent or a warm or cool bias.
  • Brand color used as text color everywhere. Mid-lightness brand hues often fail 4.5:1. Fix: use a darker ramp step for text (--color-accent-fg) and the brand step for fills.
  • Light gray text. #999 on white is 2.85:1. Fix: --color-fg-muted (4.5:1+) for secondary text; --color-fg-subtle only for decorative, non-essential text (placeholders use --color-fg-muted).
  • Status by color alone. Fix: icon plus text plus color.
  • Inverted dark mode. Pure black canvas, pure white text, same saturated accents: glare and vibration. Fix: dark mode.
  • Too many accents. Five competing hues flatten hierarchy. Fix: one accent; others become neutrals or data colors.
  • Hex-to-CMYK by formula. Fix: convert with the print profile, proof, and specify Pantone where the brand must match.

References

  • Palette architecture: roles and the kit token mapping, 11-13 step ramps, accent count, status colors, proportions, brand vs UI color.
  • OKLCH ramps: why OKLCH, the ramp algorithm, gamut mapping, and ../../library/scripts/color_tools.py usage.
  • Contrast: WCAG 2.2 math and thresholds, APCA as advisory, non-text and focus, testing.
  • Dark mode: elevation by lightness, desaturation, off-black and off-white, borders, images.
  • Meaning and culture: associations by context and market, never color alone.
  • Brand color: category mapping, ownable hues, avoiding the generic default, Pantone, CMYK and RGB specs.
  • Atlas: python3 ../../library/scripts/atlas.py query "..." and python3 ../../library/scripts/atlas.py show <id>; IDs in the registry.
  • Sibling skills: data-visualization for chart palettes, design-system for token builds and theming, typography for text sizes that set the contrast threshold, print-packaging-social for prepress.
  • Deep process for client projects: design-color-system with palette craft and proofing and maintenance; record decisions in the color decision template; worked reasoning in color in use.

Catalog IDs covered: color

04 同仓库更多 Skills

newmindsgroup
newmindsgroup/visual-design-kit

web-design

Load before planning, designing, auditing or launching any website, landing page, cinematic hero, microsite or above-the-fold layout, or a prompt for Lovable, Bolt or v0 to build one, including regulated (prescription drug, health, finance), multilingual or market-specific launches. Covers page types (home, product, pricing, about, blog), conversion and ethical persuasion, responsive and fluid layouts, Core Web Vitals, SEO and structured data, localization and language-law requirements, stack and CMS choice (Astro, Next.js, WordPress, Webflow, Framer) and launch QA.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

ux-design

Load before designing or reviewing any flow, form, onboarding, navigation or content for real users, especially low-literacy, older, neurodivergent, child or other specific audiences. Covers research and usability tests, heuristic reviews, personas and jobs to be done, journeys and service blueprints, information architecture, UX laws, SUS, SEQ and HEART metrics, plain language and reading level, and microcopy and error messages.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

ui-design

Load before designing or specifying any interface screen or component, even a quick one: forms, tables, navigation, dashboards, onboarding, auth and passkeys, pricing, checkout, search and filters, settings, empty, loading and error states, notifications, consent and age-appropriate flows, and AI chat and agent interfaces, all to WCAG 2.2 AA. Also picks React UI libraries (toasts, drawers, command menus). Gives component specs, state matrices, sizes, density and keyboard and accessibility behavior.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

typography

Choose, pair, size and set type for brand, web, app, print, slides, signage and motion: modular scales, fluid clamp() type, measure, leading, tracking, OpenType features, font licensing and web font loading, and multiscript setting for CJK, Arabic, Hebrew, Devanagari and Thai. Use when selecting or pairing fonts, building a type scale or type tokens, fixing weak hierarchy or bad line breaks, setting long-form text or numbers, loading web fonts, or localizing type for another market.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

reference-reconstruction

Reconstructs the observed interface and behavior of a supplied app, site or screenshot set using a screen index, measurements, state and command maps, reusable components and matched-condition comparisons. Use when asked to reverse engineer, clone, reproduce or match a reference product, then adapt it if requested.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

print-packaging-social

Load before any print, packaging, label, barcode, social, out-of-home, slide-deck or email design question, even a quick one, and before making finished collateral files. Specs: bleed, safe zones, 300ppi, CMYK, PDF/X preflight, dielines, barcodes, social formats with 9:16 safe areas, billboard letter heights, 16:9 decks, bulletproof email. Renders token-driven brand books, sell sheets, cards, carousels and decks to PDF and PNG with a Canva editable twin.

★ 2