Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Direction and intent for frontend design. Use with design when defining purpose, tone, domain, color world, and review bar; includes cross-domain lens from cinema, architecture, marketing, UX, automotive, industrial design.
.claude/skills/bilal140202-design-thinking/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 188% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 213% | 0% |
| case-14 | ✓→✗ | ▼ Worse | 234% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 115% | 0% |
Apply with the design skill on every design task. Do this before coding.
Commit to one bold direction. Intentionality over intensity.
High-impact devices — full-bleed media, dramatic motion, oversized type, a saturated accent, parallax — lose their force the moment they become the default. If every section is a full-screen video, none of them land. Choose 1–2 moments to go big (the hero, the payoff) and make everything else quiet and contained, so the big moments read as deliberate punctuation rather than noise. Restraint is what gives the rare full-bleed beat its hit; wall-to-wall intensity reads as flat. This is the luxury of reduction — a single bold gesture in generous negative space beats maximalism. Applies equally to scale, color (one accent, not five), and motion (one signature move, not constant animation).
Ask: "Would I put my name on this?" Design-lead review means: not "does it work?" but "is this unmistakably intentional and crafted?"
Non-negotiable craft invariants (a single failure fails the review):
document.fonts.ready and fade in. See design-system → Never let fonts pop in.When defining direction, choose which disciplines fit the product and apply 2–3 principles from each. Examples: dashboard → UX + automotive; brand campaign → cinema + marketing; app → UX + industrial design.
prefers-reduced-motion for public/multi-user projects; personal projects may ship full motion.)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→fail | 10,665 | 32,503 | +205% | 1 | 1 | 0% | 2,256 | 7,539 | +234% | 0 | 0 | — |
case-04 | fail→pass | 16,698 | 31,890 | +91% | 1 | 1 | 0% | 2,627 | 7,559 | +188% | 0 | 0 | — |
case-05 | pass→pass | 13,178 | 31,626 | +140% | 1 | 1 | 0% | 2,126 | 7,557 | +255% | 0 | 0 | — |
case-01 | fail→pass | 28,395 | 28,855 | +2% | 1 | 1 | 0% | 4,444 | 5,741 | +29% | 0 | 0 | — |
case-02 | pass→pass | 18,243 | 28,601 | +57% | 1 | 1 | 0% | 2,728 | 7,556 | +177% | 0 | 0 | — |
case-03 | pass→pass | 16,259 | 29,297 | +80% | 1 | 1 | 0% | 2,741 | 7,558 | +176% | 0 | 0 | — |
case-06 | pass→pass | 17,486 | 33,978 | +94% | 1 | 1 | 0% | 2,681 | 7,540 | +181% | 0 | 0 | — |
case-07 | pass→pass | 21,506 | 35,115 | +63% | 1 | 1 | 0% | 3,051 | 7,528 | +147% | 0 | 0 | — |
case-08 | fail→pass | 16,463 | 31,238 | +90% | 1 | 1 | 0% | 2,405 | 7,535 | +213% | 0 | 0 | — |
case-09 | pass→pass | 26,061 | 29,229 | +12% | 1 | 1 | 0% | 4,413 | 7,526 | +71% | 0 | 0 | — |
case-10 | pass→pass | 18,860 | 30,953 | +64% | 1 | 1 | 0% | 2,804 | 7,527 | +168% | 0 | 0 | — |
case-11 | fail→fail | 15,052 | 36,557 | +143% | 1 | 1 | 0% | 2,476 | 7,544 | +205% | 0 | 0 | — |
case-12 | pass→fail | 18,483 | 31,846 | +72% | 1 | 1 | 0% | 3,505 | 7,538 | +115% | 0 | 0 | — |
case-13 | pass→pass | 5,599 | 29,345 | +424% | 1 | 1 | 0% | 1,051 | 7,536 | +617% | 0 | 0 | — |
case-15 | pass→pass | 18,724 | 28,555 | +53% | 1 | 1 | 0% | 3,902 | 7,530 | +93% | 0 | 0 | — |
case-16 | pass→pass | 16,368 | 29,773 | +82% | 1 | 1 | 0% | 2,671 | 7,545 | +182% | 0 | 0 | — |
case-17 | pass→pass | 11,593 | 14,081 | +21% | 1 | 1 | 0% | 1,856 | 3,488 | +88% | 0 | 0 | — |
case-18 | fail→fail | 17,583 | 33,310 | +89% | 1 | 1 | 0% | 2,645 | 7,559 | +186% | 0 | 0 | — |
case-19 | pass→pass | 14,212 | 30,133 | +112% | 1 | 1 | 0% | 2,011 | 7,543 | +275% | 0 | 0 | — |
case-20 | pass→pass | 15,050 | 30,888 | +105% | 1 | 1 | 0% | 2,621 | 7,534 | +187% | 0 | 0 | — |
case-21 | pass→pass | 13,718 | 34,460 | +151% | 1 | 1 | 0% | 2,184 | 7,547 | +246% | 0 | 0 | — |
case-22 | pass→pass | 13,247 | 32,936 | +149% | 1 | 1 | 0% | 2,036 | 7,541 | +270% | 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 +5 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
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.