Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Open Design's brand-story deck for partners and press: why it exists, what it believes, and where design is going. Built as a decision-grade marketing & GTM deck for partners, press, community.
.claude/skills/nexu-io-open-design-landing-deck/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 66% | 0% |
Sister skill to open-design-landing. Same Atelier Zero visual system (warm paper, Inter Tight + Playfair Display, italic-serif emphasis, coral dots), but paginated as a horizontal magazine-style swipe deck instead of a long scrolling page.
The navigation model is intentionally borrowed from the guizang-ppt skill — ←/→ arrow keys, wheel / swipe, ESC for the overview grid — so it feels like a print magazine flipping page by page rather than a web slide deck scrolling.
textinputs.json + ../open-design-landing/styles.css │ └──────────► scripts/compose.ts │ ▼ <out>/index.html (one viewport per slide, horizontal swipe)
out horizontally on one transformed flex track.
←/→ · ↑/↓ · PageUp/PageDown · Space ·Home/End.
flick doesn't overshoot).
Roman-numeral year, live slide counter (01 / 11).
jump. Mirrors guizang-ppt's overview UX.
duplicate assets.
| Kind | Use it for | | :---------- | :------------------------------------------------------------ | | cover | Title plate at the start. 2-column copy + collage art. | | section | Roman-numeral divider between chapters. Centered, full-bleed. | | content | Eyebrow + title + body + bullets + optional collage art. | | stats | Up to 4 large stat cells (value · label · sub-label). | | quote | Pull quote + author. Optional portrait collage on the right. | | cta | Closing pitch + 1-2 buttons. | | end | Mega italic-serif kicker word + signature footer. |
A typical 11-slide pitch:
1. cover — title plate, hero collage
2. section — "I. The problem"
3. content — about / manifesto, bullets
4. content — capabilities, bullets
5. stats — 4 numbers
6. section — "II. How it feels"
7. content — method, bullets
8. content — selected work
9. quote — customer testimonial
10. cta — primary + secondary action
11. end — mega kicker + signatureinputs.jsonStart from inputs.example.json (the Open Design pitch deck). The brand block, image strategy, and assets path mirror the sister skill — if you already filled out an open-design-landing brief, copy brand and imagery over verbatim.
For each slide, pick a kind and fill the typed fields from schema.ts. MixedText (sans-serif baseline + italic-serif emphasis spans + coral terminating dot) is the same encoding used by the sister skill — see its inputs.example.json.
This skill does not ship its own image generator or placeholder script — it shares the 16-slot library from open-design-landing. To regenerate or stub:
bash# generate via gpt-image-2 (fal.ai) FAL_KEY=... npx tsx ../open-design-landing/scripts/imagegen.ts ../open-design-landing/inputs.example.json --out=../open-design-landing/assets/ # or paper-textured SVG placeholders npx tsx ../open-design-landing/scripts/placeholder.ts ../open-design-landing/assets/
Set your deck's inputs.imagery.assets_path to wherever those PNGs live (default in the example: ../open-design-landing/assets/).
bashnpx tsx scripts/compose.ts inputs.json out/index.html
The composer reads inputs.json, loads the canonical Atelier Zero stylesheet from ../open-design-landing/styles.css, layers deck-specific rules on top (horizontal flex track, slide layouts, HUD, dot nav, ESC overview, keyboard / wheel / touch handlers), and writes one self-contained HTML file.
with chrome strip top-right showing 01 / N.
→ (or Space, or scroll-down). Smoothly slides oneviewport to the right; dot nav advances; the coral progress bar ticks forward.
End. Jumps to the final slide.Home. Returns to slide 1.Esc. Overview grid appears with scaled thumbnails;click any tile to jump and dismiss the overview.
a single column; dot nav still works; chrome strips shrink.
prefers-reduced-motion: reduce (DevTools → Rendering): pagetransitions stay snappy and don't induce motion sickness.
../open-design-landing/styles.css at compile time. Do not maintain a duplicate copy here; if Atelier Zero tokens evolve, edit them once in the sister skill.
re-render — the same 16 plates work for both surfaces.
100vh × 100vw at 1280×800 it will overflow and feel cramped. Trim copy or split into two slides.
posture is what makes this skill feel like a magazine spread; a vertical scroller would just be a long landing page.
decks are out of scope; for a multi-deck experience, render each deck separately and link from a parent index.
open-design-landing — landing page sister skill.guizang-ppt — the magazine-deck navigationpattern this skill borrows.
design-systems/atelier-zero/DESIGN.md — token spec.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 52,925 | 34,448 | -35% | 1 | 1 | 0% | 8,259 | 2,072 | -75% | 0 | 0 | — |
case-02 | fail→fail | 52,361 | 10,106 | -81% | 1 | 1 | 0% | 8,261 | 2,013 | -76% | 0 | 0 | — |
case-03 | fail→fail | 40,713 | 8,887 | -78% | 1 | 1 | 0% | 8,247 | 1,936 | -77% | 0 | 0 | — |
case-04 | fail→pass | 13,449 | 6,937 | -48% | 1 | 1 | 0% | 2,481 | 2,710 | +9% | 0 | 0 | — |
case-05 | fail→pass | 18,333 | 9,478 | -48% | 1 | 1 | 0% | 3,481 | 3,631 | +4% | 0 | 0 | — |
case-06 | fail→pass | 14,386 | 13,670 | -5% | 1 | 1 | 0% | 2,582 | 4,016 | +56% | 0 | 0 | — |
case-07 | fail→pass | 15,358 | 1,792 | -88% | 1 | 1 | 0% | 1,797 | 2,049 | +14% | 0 | 0 | — |
case-08 | fail→pass | 12,588 | 5,509 | -56% | 1 | 1 | 0% | 1,717 | 2,858 | +66% | 0 | 0 | — |
case-09 | fail→pass | 6,382 | 3,341 | -48% | 1 | 1 | 0% | 1,071 | 2,149 | +101% | 0 | 0 | — |
case-10 | pass→pass | 10,549 | 3,109 | -71% | 1 | 1 | 0% | 1,624 | 2,048 | +26% | 0 | 0 | — |
case-11 | fail→pass | 14,491 | 3,455 | -76% | 1 | 1 | 0% | 2,404 | 2,240 | -7% | 0 | 0 | — |
case-12 | fail→pass | 8,065 | 1,967 | -76% | 1 | 1 | 0% | 1,138 | 1,966 | +73% | 0 | 0 | — |
case-13 | fail→pass | 11,484 | 5,816 | -49% | 1 | 1 | 0% | 1,824 | 2,158 | +18% | 0 | 0 | — |
case-14 | fail→pass | 10,219 | 2,734 | -73% | 1 | 1 | 0% | 1,577 | 2,169 | +38% | 0 | 0 | — |
case-15 | fail→fail | 13,514 | 1,764 | -87% | 1 | 1 | 0% | 1,291 | 2,043 | +58% | 0 | 0 | — |
case-16 | fail→pass | 6,329 | 3,491 | -45% | 1 | 1 | 0% | 851 | 2,055 | +141% | 0 | 0 | — |
case-17 | fail→pass | 10,976 | 3,729 | -66% | 1 | 1 | 0% | 1,838 | 2,194 | +19% | 0 | 0 | — |
case-18 | pass→pass | 18,019 | 2,602 | -86% | 1 | 1 | 0% | 2,528 | 2,157 | -15% | 0 | 0 | — |
case-19 | fail→pass | 11,377 | 3,077 | -73% | 1 | 1 | 0% | 1,413 | 2,231 | +58% | 0 | 0 | — |
case-20 | fail→pass | 9,571 | 3,607 | -62% | 1 | 1 | 0% | 1,462 | 2,042 | +40% | 0 | 0 | — |
case-21 | fail→pass | 8,694 | 4,700 | -46% | 1 | 1 | 0% | 1,615 | 2,597 | +61% | 0 | 0 | — |
case-22 | fail→pass | 20,605 | 11,874 | -42% | 1 | 1 | 0% | 2,934 | 3,219 | +10% | 0 | 0 | — |
case-23 | pass→pass | 17,459 | 7,989 | -54% | 1 | 1 | 0% | 2,268 | 3,066 | +35% | 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. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +70 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is 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.