Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →6 geoscience & climate skills. Trigger: earth science data, GIS, remote sensing, climate modeling. Design: geospatial tools, satellite data processing, and environmental models.
.claude/skills/brycewang-stanford-geoscience-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -62% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -63% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | climate-modeling-guide | Climate simulation, modeling tools, and climate data analysis methods | | climate-science-guide | Climate data analysis, modeling workflows, and carbon neutrality research met... | | gis-remote-sensing-guide | GIS analysis and remote sensing workflows for geospatial research applications | | pangaea-data-api | Access earth and environmental science datasets via PANGAEA API | | satellite-remote-sensing | Satellite imagery analysis and remote sensing for earth science research | | seismology-data-guide | Earthquake data analysis, seismogram processing, and seismic research |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 28,724 | 10,794 | -62% | 1 | 1 | 0% | 4,802 | 559 | -88% | 0 | 0 | — |
case-02 | fail→fail | 19,433 | 46,492 | +139% | 1 | 1 | 0% | 3,184 | 613 | -81% | 0 | 0 | — |
case-03 | fail→pass | 16,566 | 2,882 | -83% | 1 | 1 | 0% | 2,597 | 731 | -72% | 0 | 0 | — |
case-04 | fail→fail | 12,620 | 4,052 | -68% | 1 | 1 | 0% | 2,058 | 518 | -75% | 0 | 0 | — |
case-05 | fail→pass | 13,760 | 2,515 | -82% | 1 | 1 | 0% | 2,006 | 648 | -68% | 0 | 0 | — |
case-06 | fail→pass | 13,626 | 3,394 | -75% | 1 | 1 | 0% | 2,092 | 785 | -62% | 0 | 0 | — |
case-07 | fail→pass | 13,868 | 5,144 | -63% | 1 | 1 | 0% | 2,199 | 634 | -71% | 0 | 0 | — |
case-08 | fail→fail | 13,729 | 3,108 | -77% | 1 | 1 | 0% | 2,245 | 549 | -76% | 0 | 0 | — |
case-09 | pass→fail | 6,274 | 4,609 | -27% | 1 | 1 | 0% | 962 | 554 | -42% | 0 | 0 | — |
case-10 | pass→pass | 2,912 | 6,078 | +109% | 1 | 1 | 0% | 474 | 526 | +11% | 0 | 0 | — |
case-11 | pass→pass | 3,008 | 4,638 | +54% | 1 | 1 | 0% | 492 | 968 | +97% | 0 | 0 | — |
case-12 | pass→fail | 10,354 | 3,001 | -71% | 1 | 1 | 0% | 1,841 | 426 | -77% | 0 | 0 | — |
case-13 | fail→fail | 7,290 | 3,268 | -55% | 1 | 1 | 0% | 1,236 | 431 | -65% | 0 | 0 | — |
case-14 | pass→fail | 2,868 | 5,025 | +75% | 1 | 1 | 0% | 360 | 560 | +56% | 0 | 0 | — |
case-15 | pass→pass | 5,169 | 2,782 | -46% | 1 | 1 | 0% | 778 | 703 | -10% | 0 | 0 | — |
case-16 | fail→fail | 28,173 | 5,285 | -81% | 1 | 1 | 0% | 1,283 | 585 | -54% | 0 | 0 | — |
case-17 | fail→pass | 12,010 | 2,859 | -76% | 1 | 1 | 0% | 1,863 | 691 | -63% | 0 | 0 | — |
case-18 | fail→pass | 10,347 | 6,702 | -35% | 1 | 1 | 0% | 1,573 | 935 | -41% | 0 | 0 | — |
case-19 | fail→pass | 12,605 | 5,940 | -53% | 1 | 1 | 0% | 2,109 | 842 | -60% | 0 | 0 | — |
case-20 | fail→fail | 12,829 | 6,042 | -53% | 1 | 1 | 0% | 2,027 | 661 | -67% | 0 | 0 | — |
case-21 | fail→pass | 15,235 | 4,690 | -69% | 1 | 1 | 0% | 2,098 | 994 | -53% | 0 | 0 | — |
case-22 | fail→pass | 15,998 | 6,941 | -57% | 1 | 1 | 0% | 2,260 | 1,401 | -38% | 0 | 0 | — |
case-23 | fail→pass | 15,942 | 4,772 | -70% | 1 | 1 | 0% | 2,418 | 624 | -74% | 0 | 0 | — |
case-24 | fail→fail | 16,176 | 4,381 | -73% | 1 | 1 | 0% | 2,531 | 668 | -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. 24 cases were attempted, and 13 counted toward the lift figure. The other 11 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 +29 percentage points is the difference between those two pass rates over the 13 comparable cases. 4 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.