Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Spin up 2-3 expert personas to debate any topic from different angles. Structure: Opening statements, rebuttals, cross-examination, closing arguments.
.claude/skills/onewave-ai-debate-simulator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 57% | 0% |
Spin up 2-3 expert personas to debate any topic from different angles. Structure: Opening statements, rebuttals, cross-examination, closing arguments.
You are a master debate orchestrator. Create 2-3 expert personas with distinct perspectives, credentials, and rhetorical styles. Facilitate structured debates with opening statements, rebuttals, cross-examination, and closing arguments. Each persona should cite research, use logical frameworks, and directly address opposing arguments.
markdown# Debate Simulator Output **Generated**: {timestamp} --- ## Results [Your formatted output here] --- ## Recommendations [Actionable next steps]
Trigger Phrases:
Example Request: > "Sample user request here]"
Response Approach:
Remember: Focus on delivering value quickly and clearly!
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→pass | 29,754 | 31,359 | +5% | 1 | 1 | 0% | 3,909 | 4,736 | +21% | 0 | 0 | — |
case-01 | pass→pass | 28,796 | 35,915 | +25% | 1 | 1 | 0% | 3,887 | 4,452 | +15% | 0 | 0 | — |
case-02 | pass→pass | 28,369 | 26,524 | -7% | 1 | 1 | 0% | 3,754 | 3,894 | +4% | 0 | 0 | — |
case-03 | pass→pass | 43,567 | 28,799 | -34% | 1 | 1 | 0% | 5,754 | 4,256 | -26% | 0 | 0 | — |
case-05 | pass→pass | 27,619 | 31,063 | +12% | 1 | 1 | 0% | 3,646 | 4,473 | +23% | 0 | 0 | — |
case-06 | fail→pass | 31,893 | 35,092 | +10% | 1 | 1 | 0% | 4,199 | 5,061 | +21% | 0 | 0 | — |
case-07 | pass→pass | 24,725 | 26,983 | +9% | 1 | 1 | 0% | 3,483 | 4,158 | +19% | 0 | 0 | — |
case-08 | fail→pass | 20,765 | 21,344 | +3% | 1 | 1 | 0% | 2,936 | 3,228 | +10% | 0 | 0 | — |
case-09 | pass→pass | 30,628 | 34,919 | +14% | 1 | 1 | 0% | 3,989 | 5,241 | +31% | 0 | 0 | — |
case-10 | fail→pass | 30,599 | 30,868 | +1% | 1 | 1 | 0% | 3,754 | 4,168 | +11% | 0 | 0 | — |
case-11 | pass→pass | 20,195 | 29,232 | +45% | 1 | 1 | 0% | 2,766 | 4,240 | +53% | 0 | 0 | — |
case-12 | fail→pass | 26,893 | 29,290 | +9% | 1 | 1 | 0% | 3,697 | 4,337 | +17% | 0 | 0 | — |
case-13 | pass→pass | 20,536 | 27,240 | +33% | 1 | 1 | 0% | 3,343 | 4,513 | +35% | 0 | 0 | — |
case-14 | fail→pass | 17,347 | 27,242 | +57% | 1 | 1 | 0% | 2,889 | 4,530 | +57% | 0 | 0 | — |
case-15 | pass→pass | 30,500 | 27,596 | -10% | 1 | 1 | 0% | 4,242 | 4,734 | +12% | 0 | 0 | — |
case-16 | fail→pass | 22,960 | 25,973 | +13% | 1 | 1 | 0% | 3,260 | 3,989 | +22% | 0 | 0 | — |
case-17 | pass→pass | 17,662 | 25,208 | +43% | 1 | 1 | 0% | 2,616 | 3,921 | +50% | 0 | 0 | — |
case-18 | fail→pass | 20,269 | 18,768 | -7% | 1 | 1 | 0% | 3,181 | 3,336 | +5% | 0 | 0 | — |
case-19 | pass→pass | 21,185 | 25,819 | +22% | 1 | 1 | 0% | 3,363 | 4,470 | +33% | 0 | 0 | — |
case-20 | pass→pass | 13,702 | 15,017 | +10% | 1 | 1 | 0% | 2,443 | 2,684 | +10% | 0 | 0 | — |
case-21 | pass→fail | 10,110 | 18,384 | +82% | 1 | 1 | 0% | 1,831 | 3,505 | +91% | 0 | 0 | — |
case-22 | pass→fail | 12,724 | 19,586 | +54% | 1 | 1 | 0% | 2,217 | 3,437 | +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. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.