Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Browse and select color palettes from Coolors or curated fallbacks. Use to find the perfect color palette for a design project.
.claude/skills/davepoon-color-curator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 225% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 79% | 0% |
| case-06 | ✓→✓ | = Same ✓ | 65% | 0% |
Browse, select, and apply color palettes for frontend designs.
This skill helps select the perfect color palette by:
javascripttabs_context_mcp({ createIfEmpty: true }) tabs_create_mcp() navigate({ url: "https://coolors.co/palettes/trending", tabId: tabId })
Take screenshots of trending palettes:
javascriptcomputer({ action: "screenshot", tabId: tabId })
Present to user: "Here are the trending palettes. Which one catches your eye?"
If user wants more options:
javascriptcomputer({ action: "scroll", scroll_direction: "down", tabId: tabId }) computer({ action: "screenshot", tabId: tabId })
When user chooses a palette, click to view details:
javascriptcomputer({ action: "left_click", coordinate: [x, y], tabId: tabId })
From the palette detail view, extract:
Based on user's background style preference:
| Background Style | Mapping | |-----------------|---------| | Pure white | bg: #ffffff, text: darkest color | | Off-white/warm | bg: #faf8f5, text: darkest | | Light tinted | bg: lightest from palette, text: darkest | | Dark/moody | bg: darkest from palette, text: white/#fafafa |
Create Tailwind color config:
javascripttailwind.config = { theme: { extend: { colors: { primary: '#[main-color]', secondary: '#[supporting-color]', accent: '#[pop-color]', background: '#[bg-color]', surface: '#[card-color]', text: { primary: '#[heading-color]', secondary: '#[body-color]', muted: '#[muted-color]' } } } } }
When browser tools are unavailable, use curated palettes.
references/color-theory.mdAsk the user:
"Without browser access, I can suggest palettes based on your aesthetic. Which mood fits best?"
Users can also provide:
Always verify:
| Role | Usage | Count | |------|-------|-------| | Primary | Brand, CTAs, links | 1 | | Secondary | Hover, icons, supporting | 1-2 | | Background | Page background | 1 | | Surface | Cards, modals, inputs | 1 | | Border | Dividers, outlines | 1 | | Text Primary | Headings, important text | 1 | | Text Secondary | Body, descriptions | 1 | | Text Muted | Captions, placeholders | 1 |
Provide the selected palette in this format:
markdown## Selected Color Palette ### Colors | Role | Hex | Preview | Usage | |------|-----|---------|-------| | Primary | #ff6b35 | 🟧 | CTAs, links, accents | | Background | #0a0a0a | ⬛ | Page background | | Surface | #1a1a1a | ⬛ | Cards, modals | | Text Primary | #ffffff | ⬜ | Headings, buttons | | Text Secondary | #a3a3a3 | ⬜ | Body text, descriptions | | Border | #2a2a2a | ⬛ | Dividers, outlines | ### Tailwind Config \`\`\`javascript colors: { primary: '#ff6b35', background: '#0a0a0a', surface: '#1a1a1a', text: { primary: '#ffffff', secondary: '#a3a3a3', }, border: '#2a2a2a', } \`\`\` ### CSS Variables (Alternative) \`\`\`css :root { --color-primary: #ff6b35; --color-background: #0a0a0a; --color-surface: #1a1a1a; --color-text-primary: #ffffff; --color-text-secondary: #a3a3a3; --color-border: #2a2a2a; } \`\`\`
See references/color-theory.md for:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,409 | 10,316 | -33% | 1 | 1 | 0% | 2,985 | 2,852 | -4% | 0 | 0 | — |
case-02 | fail→pass | 15,545 | 9,652 | -38% | 1 | 1 | 0% | 2,828 | 3,390 | +20% | 0 | 0 | — |
case-03 | fail→fail | 5,799 | 4,921 | -15% | 1 | 1 | 0% | 983 | 2,241 | +128% | 0 | 0 | — |
case-04 | pass→pass | 3,343 | 3,586 | +7% | 1 | 1 | 0% | 634 | 2,060 | +225% | 0 | 0 | — |
case-05 | pass→pass | 6,835 | 3,762 | -45% | 1 | 1 | 0% | 1,169 | 2,097 | +79% | 0 | 0 | — |
case-06 | pass→pass | 7,740 | 5,006 | -35% | 1 | 1 | 0% | 1,374 | 2,266 | +65% | 0 | 0 | — |
case-07 | pass→pass | 4,866 | 3,443 | -29% | 1 | 1 | 0% | 846 | 1,962 | +132% | 0 | 0 | — |
case-08 | pass→pass | 12,430 | 3,960 | -68% | 1 | 1 | 0% | 2,292 | 2,254 | -2% | 0 | 0 | — |
case-09 | pass→pass | 6,596 | 5,486 | -17% | 1 | 1 | 0% | 1,239 | 2,530 | +104% | 0 | 0 | — |
case-10 | pass→pass | 7,062 | 2,741 | -61% | 1 | 1 | 0% | 1,115 | 1,926 | +73% | 0 | 0 | — |
case-11 | pass→pass | 12,140 | 9,599 | -21% | 1 | 1 | 0% | 2,502 | 3,503 | +40% | 0 | 0 | — |
case-12 | pass→pass | 6,678 | 4,122 | -38% | 1 | 1 | 0% | 1,145 | 2,377 | +108% | 0 | 0 | — |
case-13 | fail→pass | 6,719 | 1,389 | -79% | 1 | 1 | 0% | 1,101 | 1,656 | +50% | 0 | 0 | — |
case-14 | pass→pass | 11,038 | 2,674 | -76% | 1 | 1 | 0% | 2,264 | 1,904 | -16% | 0 | 0 | — |
case-15 | pass→pass | 8,429 | 4,570 | -46% | 1 | 1 | 0% | 1,358 | 2,237 | +65% | 0 | 0 | — |
case-16 | pass→pass | 2,723 | 1,385 | -49% | 1 | 1 | 0% | 459 | 1,668 | +263% | 0 | 0 | — |
case-17 | fail→fail | 4,946 | 1,810 | -63% | 1 | 1 | 0% | 799 | 1,707 | +114% | 0 | 0 | — |
case-18 | pass→pass | 7,732 | 3,365 | -56% | 1 | 1 | 0% | 1,485 | 2,065 | +39% | 0 | 0 | — |
case-19 | fail→fail | 6,701 | 1,534 | -77% | 1 | 1 | 0% | 1,159 | 1,701 | +47% | 0 | 0 | — |
case-20 | pass→pass | 12,739 | 10,268 | -19% | 1 | 1 | 0% | 2,855 | 3,834 | +34% | 0 | 0 | — |
case-21 | pass→pass | 9,928 | 7,585 | -24% | 1 | 1 | 0% | 1,974 | 3,072 | +56% | 0 | 0 | — |
case-22 | pass→pass | 5,394 | 4,255 | -21% | 1 | 1 | 0% | 992 | 2,329 | +135% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.