Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →9 economics skills. Trigger: economic modeling, policy analysis, macroeconomic data, FRED. Design: theory plus empirical methods with standard economics databases.
.claude/skills/brycewang-stanford-economics-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-01 | ✓→✗ | ▼ Worse | -73% | 0% |
| case-07 | ✓→✗ | ▼ Worse | -70% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | behavioral-economics-guide | Behavioral economics research methods and key frameworks | | development-economics-guide | Apply development economics research methods and data sources | | fred-api | Federal Reserve Economic Data API for US economic indicators | | imf-data-api-guide | Retrieve IMF economic indicators, exchange rates, and country data | | nber-working-papers-api | Access NBER working papers and economic research datasets | | post-labor-economics | Post-labor economies with automation, UBI, and wealth distribution | | pricing-psychology-guide | Behavioral economics in pricing strategies and consumer decisions | | repec-economics-api | Access 4M+ economics working papers and articles via RePEc API | | world-bank-data-guide | Access World Bank development indicators and country statistics |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 14,503 | 70,449 | +386% | 1 | 1 | 0% | 2,117 | 753 | -64% | 0 | 0 | — |
case-03 | fail→pass | 13,683 | 3,919 | -71% | 1 | 1 | 0% | 2,014 | 1,011 | -50% | 0 | 0 | — |
case-01 | pass→fail | 11,788 | 5,161 | -56% | 1 | 1 | 0% | 1,987 | 530 | -73% | 0 | 0 | — |
case-04 | fail→pass | 9,322 | 3,211 | -66% | 1 | 1 | 0% | 1,544 | 825 | -47% | 0 | 0 | — |
case-05 | fail→fail | 8,622 | 5,751 | -33% | 1 | 1 | 0% | 1,358 | 767 | -44% | 0 | 0 | — |
case-06 | fail→fail | 18,312 | 19,583 | +7% | 1 | 1 | 0% | 2,575 | 546 | -79% | 0 | 0 | — |
case-07 | pass→fail | 15,900 | 4,073 | -74% | 1 | 1 | 0% | 2,185 | 660 | -70% | 0 | 0 | — |
case-08 | fail→fail | 16,733 | 4,989 | -70% | 1 | 1 | 0% | 2,530 | 692 | -73% | 0 | 0 | — |
case-09 | fail→fail | 7,863 | 4,322 | -45% | 1 | 1 | 0% | 1,244 | 740 | -41% | 0 | 0 | — |
case-10 | pass→fail | 6,876 | 4,290 | -38% | 1 | 1 | 0% | 895 | 587 | -34% | 0 | 0 | — |
case-11 | pass→pass | 8,086 | 5,939 | -27% | 1 | 1 | 0% | 1,192 | 868 | -27% | 0 | 0 | — |
case-12 | pass→pass | 6,096 | 3,135 | -49% | 1 | 1 | 0% | 856 | 803 | -6% | 0 | 0 | — |
case-13 | fail→fail | 14,150 | 4,729 | -67% | 1 | 1 | 0% | 2,135 | 748 | -65% | 0 | 0 | — |
case-14 | fail→fail | 11,248 | 5,244 | -53% | 1 | 1 | 0% | 1,762 | 845 | -52% | 0 | 0 | — |
case-15 | pass→pass | 17,604 | 3,839 | -78% | 1 | 1 | 0% | 2,455 | 993 | -60% | 0 | 0 | — |
case-16 | pass→fail | 17,152 | 10,905 | -36% | 1 | 1 | 0% | 3,128 | 789 | -75% | 0 | 0 | — |
case-17 | pass→fail | 6,070 | 4,411 | -27% | 1 | 1 | 0% | 973 | 661 | -32% | 0 | 0 | — |
case-18 | pass→fail | 13,926 | 6,178 | -56% | 1 | 1 | 0% | 2,212 | 782 | -65% | 0 | 0 | — |
case-19 | fail→pass | 10,155 | 2,276 | -78% | 1 | 1 | 0% | 1,542 | 694 | -55% | 0 | 0 | — |
case-20 | pass→pass | 9,461 | 5,992 | -37% | 1 | 1 | 0% | 1,555 | 1,418 | -9% | 0 | 0 | — |
case-21 | pass→pass | 6,096 | 3,550 | -42% | 1 | 1 | 0% | 1,018 | 950 | -7% | 0 | 0 | — |
case-22 | pass→pass | 14,592 | 11,415 | -22% | 1 | 1 | 0% | 2,997 | 2,915 | -3% | 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 9 counted toward the lift figure. The other 13 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 -14 percentage points is the difference between those two pass rates over the 9 comparable cases. 6 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.