Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →6 social science skills. Trigger: survey research, social networks, psychology, behavioral studies. Design: quantitative and qualitative methods for social science research.
.claude/skills/brycewang-stanford-social-science-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✓→✗ | ▼ Worse | -40% | 0% |
| case-02 | ✓→✗ | ▼ Worse | -42% | 0% |
| case-01 | ✓→✗ | ▼ Worse | -77% | 0% |
| case-04 | ✓→✗ | ▼ Worse | -59% | 0% |
| case-05 | ✓→✗ | ▼ Worse | -56% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | ipums-microdata-api | Access harmonized census and survey microdata via the IPUMS API | | network-analysis-guide | Social network analysis methods, metrics, and visualization tools | | psychology-research-guide | Psychological research methods, experimental design, and analysis | | social-research-methods | Core methods for empirical social science research including surveys and expe... | | sociology-research-methods | Sociological research methods from observation to quantitative analysis | | survey-research-guide | Design, deploy, and analyze surveys for social science and organizational res... |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→fail | 5,280 | 4,136 | -22% | 1 | 1 | 0% | 850 | 508 | -40% | 0 | 0 | — |
case-02 | pass→fail | 5,051 | 3,667 | -27% | 1 | 1 | 0% | 915 | 533 | -42% | 0 | 0 | — |
case-01 | pass→fail | 13,117 | 7,035 | -46% | 1 | 1 | 0% | 2,242 | 506 | -77% | 0 | 0 | — |
case-04 | pass→fail | 7,023 | 3,324 | -53% | 1 | 1 | 0% | 1,152 | 468 | -59% | 0 | 0 | — |
case-05 | pass→fail | 6,749 | 3,992 | -41% | 1 | 1 | 0% | 1,198 | 528 | -56% | 0 | 0 | — |
case-06 | pass→fail | 17,261 | 3,662 | -79% | 1 | 1 | 0% | 2,684 | 471 | -82% | 0 | 0 | — |
case-07 | pass→fail | 6,715 | 7,196 | +7% | 1 | 1 | 0% | 1,004 | 703 | -30% | 0 | 0 | — |
case-08 | pass→fail | 14,586 | 5,151 | -65% | 1 | 1 | 0% | 2,402 | 563 | -77% | 0 | 0 | — |
case-09 | pass→fail | 12,094 | 5,034 | -58% | 1 | 1 | 0% | 1,910 | 614 | -68% | 0 | 0 | — |
case-10 | pass→fail | 12,984 | 6,714 | -48% | 1 | 1 | 0% | 1,982 | 590 | -70% | 0 | 0 | — |
case-11 | pass→pass | 6,351 | 5,283 | -17% | 1 | 1 | 0% | 1,197 | 1,200 | +0% | 0 | 0 | — |
case-12 | pass→pass | 4,113 | 8,556 | +108% | 1 | 1 | 0% | 567 | 1,229 | +117% | 0 | 0 | — |
case-13 | pass→fail | 8,139 | 4,727 | -42% | 1 | 1 | 0% | 1,222 | 673 | -45% | 0 | 0 | — |
case-14 | pass→pass | 7,038 | 6,865 | -2% | 1 | 1 | 0% | 1,133 | 1,380 | +22% | 0 | 0 | — |
case-15 | pass→fail | 11,642 | 3,918 | -66% | 1 | 1 | 0% | 1,810 | 462 | -74% | 0 | 0 | — |
case-16 | pass→fail | 5,473 | 6,031 | +10% | 1 | 1 | 0% | 881 | 735 | -17% | 0 | 0 | — |
case-17 | pass→fail | 4,436 | 3,459 | -22% | 1 | 1 | 0% | 581 | 515 | -11% | 0 | 0 | — |
case-18 | pass→fail | 14,825 | 13,116 | -12% | 1 | 1 | 0% | 2,102 | 579 | -72% | 0 | 0 | — |
case-19 | pass→fail | 5,876 | 6,223 | +6% | 1 | 1 | 0% | 852 | 534 | -37% | 0 | 0 | — |
case-20 | pass→fail | 8,592 | 5,875 | -32% | 1 | 1 | 0% | 1,254 | 593 | -53% | 0 | 0 | — |
case-21 | pass→pass | 7,388 | 8,509 | +15% | 1 | 1 | 0% | 1,280 | 1,578 | +23% | 0 | 0 | — |
case-22 | pass→pass | 5,324 | 7,943 | +49% | 1 | 1 | 0% | 909 | 1,542 | +70% | 0 | 0 | — |
case-23 | pass→fail | 9,917 | 25,583 | +158% | 1 | 1 | 0% | 1,586 | 614 | -61% | 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 5 counted toward the lift figure. The other 18 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. A headline lift is not published for this run.
Other measured skills in the registry, with their headline benchmark lift.