Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A lab-meeting deck on gut-microbiome links to sleep quality — the design, the results, the caveats, and the next experiment. Built as a decision-grade academic research deck for lab group, PI.
.claude/skills/nexu-io-kami-deck/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-18 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 58% | 0% |
Sister skill to kami-landing. Produces a single self-contained HTML file: a horizontal magazine-style swipe deck in the kami (紙 / 纸) design system — print rhythm, ink-blue accent, serif at one weight, no italic, no cool grays.
The navigation model is intentionally borrowed from the guizang-ppt skill — ←/→ arrow keys, wheel / swipe, ESC for the overview grid. The aesthetic stays kami: parchment content slides, ink-blue cover and chapter slides, serif everywhere.
> Design system source of truth: > design-systems/kami/DESIGN.md. > Read it before shipping. Tokens, type rules, and forbidden colors > all live there. Slide-specific scale ratios (macro × 1.6, > letter-spacing × 0.6 vs. print) are documented in §3 "Hierarchy" > and §5 "Layout Principles · Slides".
horizontally on one transformed flex track.
(#1B365D) with ivory text — the only place dark theme is used.
(#f5f4ed) with serif at weight 500.
counter (01 / 09).
nav are ink-blue too.
:root viathe language parameter.
| Kind | Background | Use it for | | :---------- | :--------- | :-------------------------------------------------------- | | cover | ink-blue | Title plate at the start. Centered serif title + tagline. | | chapter | ink-blue | Roman/Arabic numeral chapter divider. | | content | parchment | Section number + title + body + optional bullets. | | stats | parchment | 3-4 metric cells (value · label · sub). | | quote | parchment | Pull quote with ink-blue left rule + author signature. | | cta | parchment | Closing pitch + 1-2 buttons. | | end | ink-blue | Mega serif kicker word + colophon footer. |
A typical 11-slide deck:
1. cover — ink-blue title plate
2. chapter — "01 / Why now"
3. content — manifesto
4. content — capabilities + bullets
5. stats — 4 numbers
6. chapter — "02 / How it feels"
7. content — method
8. content — selected work
9. quote — testimonial
10. cta — primary action
11. end — ink-blue kickerAsk in two rounds (don't dump the whole list at once):
Same as kami-landing: EN → Charter, zh-CN → TsangerJinKai02 / Source Han Serif, ja → YuMincho. JA also overrides --olive to #4d4c48 because YuMincho strokes are thinner.
index.htmlOutput a single file with all CSS inline. Mirror the structure of example.html. Use only the tokens from design-systems/kami/DESIGN.md.
The runtime script (keyboard / wheel / touch nav, dot indicator, progress bar, ESC overview) should match the model documented in open-design-landing-deck/scripts/compose.ts. Do not reuse the open-design-landing-deck CSS; the visual language is different.
(#1B365D) with ivory text. All other slides are on parchment.
font-variant-numeric: tabular-nums.
→ / Space / scroll. Smoothly slides one viewportto the right; dot nav advances; the ink-blue progress bar ticks forward.
Esc. Overview grid appears with scaled thumbnails.single column; dot nav still works.
pick nothing.
rgba() for tag fills; pre-blend over parchmentand use solid hex from the table in design-systems/kami/DESIGN.md §2.
visual system). Kami is gradient-free, image-light, and hierarchy is carried by type.
kami-landing — long-form one-pager sister skill.design-systems/kami/DESIGN.md — token spec.open-design-landing-deck — samehorizontal swipe nav model, different visual language (Atelier Zero).
tw93/kami — originalClaude skill (MIT). Kami's slides.py template documents the macro × 1.6 / micro × 0.6 ratios this skill applies.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→fail | 29,882 | 49,539 | +66% | 1 | 1 | 0% | 4,621 | 9,832 | +113% | 0 | 0 | — |
case-18 | fail→pass | 14,211 | 15,049 | +6% | 1 | 1 | 0% | 2,221 | 3,803 | +71% | 0 | 0 | — |
case-03 | fail→fail | 8,982 | 5,209 | -42% | 1 | 1 | 0% | 1,294 | 2,410 | +86% | 0 | 0 | — |
case-01 | fail→pass | 11,454 | 8,909 | -22% | 1 | 1 | 0% | 1,805 | 2,514 | +39% | 0 | 0 | — |
case-02 | fail→pass | 10,385 | 7,343 | -29% | 1 | 1 | 0% | 1,762 | 2,758 | +57% | 0 | 0 | — |
case-05 | pass→fail | 32,881 | 25,537 | -22% | 1 | 1 | 0% | 6,867 | 6,591 | -4% | 0 | 0 | — |
case-06 | pass→pass | 30,219 | 49,792 | +65% | 1 | 1 | 0% | 6,142 | 9,817 | +60% | 0 | 0 | — |
case-07 | fail→pass | 20,133 | 17,750 | -12% | 1 | 1 | 0% | 3,463 | 3,606 | +4% | 0 | 0 | — |
case-08 | pass→pass | 12,904 | 11,706 | -9% | 1 | 1 | 0% | 2,053 | 3,529 | +72% | 0 | 0 | — |
case-09 | fail→pass | 14,931 | 11,121 | -26% | 1 | 1 | 0% | 1,955 | 3,087 | +58% | 0 | 0 | — |
case-10 | fail→pass | 18,841 | 5,917 | -69% | 1 | 1 | 0% | 2,622 | 2,565 | -2% | 0 | 0 | — |
case-11 | fail→pass | 13,448 | 4,452 | -67% | 1 | 1 | 0% | 2,100 | 2,284 | +9% | 0 | 0 | — |
case-12 | fail→pass | 15,837 | 2,471 | -84% | 1 | 1 | 0% | 1,719 | 1,920 | +12% | 0 | 0 | — |
case-13 | pass→pass | 12,687 | 9,067 | -29% | 1 | 1 | 0% | 2,199 | 3,149 | +43% | 0 | 0 | — |
case-14 | fail→pass | 8,750 | 3,132 | -64% | 1 | 1 | 0% | 1,111 | 1,975 | +78% | 0 | 0 | — |
case-15 | fail→pass | 18,100 | 10,845 | -40% | 1 | 1 | 0% | 2,452 | 3,065 | +25% | 0 | 0 | — |
case-16 | fail→pass | 16,514 | 9,563 | -42% | 1 | 1 | 0% | 2,760 | 2,830 | +3% | 0 | 0 | — |
case-17 | pass→pass | 12,047 | 5,750 | -52% | 1 | 1 | 0% | 1,866 | 2,312 | +24% | 0 | 0 | — |
case-19 | fail→pass | 13,008 | 3,864 | -70% | 1 | 1 | 0% | 1,730 | 2,199 | +27% | 0 | 0 | — |
case-20 | fail→pass | 12,633 | 6,642 | -47% | 1 | 1 | 0% | 1,748 | 2,494 | +43% | 0 | 0 | — |
case-21 | fail→pass | 13,023 | 12,472 | -4% | 1 | 1 | 0% | 2,304 | 4,062 | +76% | 0 | 0 | — |
case-22 | pass→pass | 14,137 | 15,004 | +6% | 1 | 1 | 0% | 2,196 | 3,590 | +63% | 0 | 0 | — |
case-23 | fail→pass | 7,447 | 3,012 | -60% | 1 | 1 | 0% | 1,132 | 1,968 | +74% | 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. The headline lift of +57 percentage points is the difference between those two pass rates over the 23 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.