Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generates an HTML gallery of all project figures with captions and source notebooks. Use when reviewing figures.
.claude/skills/brycewang-stanford-figures-gallery/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 152% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -49% | 0% |
Create an HTML gallery page displaying all project figures with their captions and source notebooks.
images/ for all image files:.png, .jpg, .jpeg, .svg, .pdf_quarto.yml) for cells with fig- labels:fig-sample, fig-event-study)#| fig-cap: or *| fig-cap: directivesimages/:{{< embed notebooks/<name>.ipynb#<label> >}}_manuscript/figures-gallery.html (the _manuscript/ directory is gitignored, so this output won't be committed).images/open _manuscript/figures-gallery.htmlimages/ does not exist or is empty, note it and continue with notebook figures.fig- labeled cells, note it and continue with images/ files.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,762 | 5,021 | -13% | 1 | 1 | 0% | 853 | 562 | -34% | 0 | 0 | — |
case-02 | fail→fail | 21,287 | 10,651 | -50% | 1 | 1 | 0% | 4,951 | 1,174 | -76% | 0 | 0 | — |
case-03 | fail→fail | 4,526 | 6,687 | +48% | 1 | 1 | 0% | 164 | 768 | +368% | 0 | 0 | — |
case-04 | fail→pass | 11,670 | 18,532 | +59% | 1 | 1 | 0% | 1,864 | 4,699 | +152% | 0 | 0 | — |
case-05 | fail→fail | 8,340 | 6,774 | -19% | 1 | 1 | 0% | 1,412 | 1,326 | -6% | 0 | 0 | — |
case-06 | pass→pass | 3,055 | 1,998 | -35% | 1 | 1 | 0% | 571 | 797 | +40% | 0 | 0 | — |
case-07 | fail→pass | 9,100 | 2,670 | -71% | 1 | 1 | 0% | 1,669 | 904 | -46% | 0 | 0 | — |
case-08 | pass→pass | 13,836 | 11,401 | -18% | 1 | 1 | 0% | 2,469 | 2,558 | +4% | 0 | 0 | — |
case-09 | pass→pass | 6,812 | 2,052 | -70% | 1 | 1 | 0% | 1,203 | 813 | -32% | 0 | 0 | — |
case-10 | fail→pass | 13,085 | 2,418 | -82% | 1 | 1 | 0% | 2,386 | 854 | -64% | 0 | 0 | — |
case-11 | fail→pass | 4,396 | 1,501 | -66% | 1 | 1 | 0% | 751 | 682 | -9% | 0 | 0 | — |
case-12 | fail→pass | 8,822 | 2,275 | -74% | 1 | 1 | 0% | 1,558 | 799 | -49% | 0 | 0 | — |
case-13 | pass→pass | 11,066 | 1,778 | -84% | 1 | 1 | 0% | 1,836 | 739 | -60% | 0 | 0 | — |
case-14 | pass→pass | 7,036 | 1,514 | -78% | 1 | 1 | 0% | 1,181 | 696 | -41% | 0 | 0 | — |
case-15 | pass→pass | 8,348 | 1,284 | -85% | 1 | 1 | 0% | 1,393 | 628 | -55% | 0 | 0 | — |
case-16 | pass→pass | 9,899 | 1,734 | -82% | 1 | 1 | 0% | 1,755 | 698 | -60% | 0 | 0 | — |
case-17 | fail→pass | 12,400 | 5,985 | -52% | 1 | 1 | 0% | 2,392 | 1,581 | -34% | 0 | 0 | — |
case-18 | fail→pass | 12,978 | 1,352 | -90% | 1 | 1 | 0% | 827 | 660 | -20% | 0 | 0 | — |
case-19 | pass→pass | 14,072 | 8,746 | -38% | 1 | 1 | 0% | 2,307 | 2,296 | -0% | 0 | 0 | — |
case-20 | pass→pass | 10,715 | 5,638 | -47% | 1 | 1 | 0% | 1,999 | 1,466 | -27% | 0 | 0 | — |
case-21 | pass→pass | 9,632 | 8,635 | -10% | 1 | 1 | 0% | 1,982 | 2,208 | +11% | 0 | 0 | — |
case-22 | pass→pass | 6,679 | 4,613 | -31% | 1 | 1 | 0% | 1,353 | 1,270 | -6% | 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 18 counted toward the lift figure. The other 4 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 +32 percentage points is the difference between those two pass rates over the 18 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.