Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Creates a Quarto revealjs slide deck in slides/ with the project style guide. Use when a presentation is needed.
.claude/skills/brycewang-stanford-new-slide-deck/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 134% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 100% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-07 | ✓→✗ | ▼ Worse | 4% | 0% |
| case-11 | ✓→✗ | ▼ Worse | -62% | 0% |
Create a new Quarto revealjs slide deck in slides/ with the project's style guide applied.
$ARGUMENTS — the presentation title and optional subtitle (e.g., "Regional Disparities in GDP" or "Job Market Talk: Regional Disparities")slides/ for existing files to avoid name collisions. Generate a filename using the naming convention from slides/README.md (e.g., analysis-results.qmd or 01-seminar-talk.qmd)..qmd file in slides/ with this YAML front matter (from the style guide in slides/README.md):yaml --- title: "<title>" subtitle: "<subtitle if provided>" author: "<from index.qmd author field, or [FILL: Author name]>" date: today format: revealjs: theme: simple slide-number: true css: | .reveal h1, .reveal h2, .reveal h3 { color: #2874A6; } .reveal strong { color: #229954; } ---
markdown ## Motivation
## Related Literature
## Data
## Empirical Strategy
## Main Results
## Robustness
## Conclusion
## Thank You {.center}
Contact: email]
bash quarto render slides/<filename>.qmd
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→fail | 23,158 | 5,620 | -76% | 1 | 1 | 0% | 4,680 | 767 | -84% | 0 | 0 | — |
case-01 | fail→fail | 17,181 | 5,479 | -68% | 1 | 1 | 0% | 2,880 | 885 | -69% | 0 | 0 | — |
case-07 | pass→fail | 4,974 | 4,432 | -11% | 1 | 1 | 0% | 782 | 810 | +4% | 0 | 0 | — |
case-02 | fail→fail | 18,259 | 5,093 | -72% | 1 | 1 | 0% | 3,012 | 828 | -73% | 0 | 0 | — |
case-03 | fail→fail | 14,235 | 5,286 | -63% | 1 | 1 | 0% | 2,727 | 847 | -69% | 0 | 0 | — |
case-04 | fail→fail | 26,111 | 5,759 | -78% | 1 | 1 | 0% | 4,834 | 896 | -81% | 0 | 0 | — |
case-05 | fail→fail | 22,250 | 4,135 | -81% | 1 | 1 | 0% | 4,608 | 845 | -82% | 0 | 0 | — |
case-08 | fail→pass | 4,828 | 7,599 | +57% | 1 | 1 | 0% | 840 | 1,965 | +134% | 0 | 0 | — |
case-09 | fail→fail | 5,826 | 4,188 | -28% | 1 | 1 | 0% | 963 | 789 | -18% | 0 | 0 | — |
case-10 | fail→fail | 15,031 | 6,040 | -60% | 1 | 1 | 0% | 2,420 | 914 | -62% | 0 | 0 | — |
case-11 | pass→fail | 12,331 | 5,651 | -54% | 1 | 1 | 0% | 2,245 | 844 | -62% | 0 | 0 | — |
case-12 | pass→fail | 18,319 | 4,407 | -76% | 1 | 1 | 0% | 3,487 | 719 | -79% | 0 | 0 | — |
case-13 | fail→fail | 30,728 | 5,142 | -83% | 1 | 1 | 0% | 5,639 | 840 | -85% | 0 | 0 | — |
case-14 | fail→fail | 28,393 | 4,800 | -83% | 1 | 1 | 0% | 6,162 | 745 | -88% | 0 | 0 | — |
case-15 | fail→fail | 5,051 | 3,462 | -31% | 1 | 1 | 0% | 955 | 1,201 | +26% | 0 | 0 | — |
case-16 | fail→pass | 2,217 | 1,294 | -42% | 1 | 1 | 0% | 368 | 736 | +100% | 0 | 0 | — |
case-17 | fail→fail | 4,284 | 5,013 | +17% | 1 | 1 | 0% | 766 | 869 | +13% | 0 | 0 | — |
case-18 | fail→fail | 12,309 | 4,399 | -64% | 1 | 1 | 0% | 2,071 | 778 | -62% | 0 | 0 | — |
case-19 | pass→fail | 6,163 | 4,881 | -21% | 1 | 1 | 0% | 1,045 | 792 | -24% | 0 | 0 | — |
case-20 | pass→pass | 5,241 | 2,674 | -49% | 1 | 1 | 0% | 1,023 | 999 | -2% | 0 | 0 | — |
case-21 | fail→pass | 8,673 | 4,165 | -52% | 1 | 1 | 0% | 1,531 | 1,392 | -9% | 0 | 0 | — |
case-22 | pass→fail | 10,490 | 3,556 | -66% | 1 | 1 | 0% | 1,639 | 740 | -55% | 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, and 5 counted toward the lift figure. The other 17 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 -9 percentage points is the difference between those two pass rates over the 5 comparable cases. 7 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.