Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Design worst possible solution → extract insights → invert. Deliberate failure design as creative catalyst.
.claude/skills/yogsoth-ai-worst-method-inversion/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 160% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 106% | 0% |
Design worst possible solution → extract insights → invert.
| Resource | Target | Current | % | |----------|--------|---------|---| | web-search | 15 | 0 | 0% | | web-research | 5 | 0 | 0% | | paper-overview | 15 | 0 | 0% | | paper-search | 10 | 0 | 0% | | paper-research | 3 | 0 | 0% |
Cannot exit strategy until ≥80% of each budget line is consumed OR yield targets are met with justification for remaining budget.
| Tactic | Role | |--------|------| | inversion-protocol | Reverse → extract → build constructive alternatives |
| SOP | Role | |-----|------| | worst-case-design | Design the worst possible solution in detail | | inversion-extraction | Extract why it's bad → what the opposite implies | | constructive-rebellion | Build constructive solutions from extracted insights | | destruction-synthesis | Synthesize inversion outputs |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | inversion-protocol | Reverse statements → extract insights → build constructive alternatives. Systematic inversion pipeline from negation to innovation. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | constructive-rebellion | Build constructive alternatives from destructive negation. Transform violated assumptions into viable innovation directions. | | destruction-synthesis | Synthesize all assumption destruction outputs into structured destructive innovation report. | | inversion-extraction | Extract constructive insights from worst solutions. Transform failure analysis into innovation directions. | | worst-case-design | Design the worst possible solution. Deliberate failure engineering to reveal hidden constraints and inversion opportunities. |
<!-- END available-tables (generated) -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 47,081 | 40,655 | -14% | 1 | 1 | 0% | 6,700 | 4,352 | -35% | 0 | 0 | — |
case-02 | fail→fail | 58,501 | 78,967 | +35% | 1 | 1 | 0% | 8,261 | 8,829 | +7% | 0 | 0 | — |
case-03 | fail→fail | 28,791 | 69,802 | +142% | 1 | 1 | 0% | 3,753 | 8,825 | +135% | 0 | 0 | — |
case-04 | pass→fail | 16,617 | 45,680 | +175% | 1 | 1 | 0% | 2,638 | 6,854 | +160% | 0 | 0 | — |
case-05 | pass→pass | 22,757 | 57,487 | +153% | 1 | 1 | 0% | 3,276 | 8,801 | +169% | 0 | 0 | — |
case-06 | pass→fail | 25,953 | 51,064 | +97% | 1 | 1 | 0% | 4,269 | 8,792 | +106% | 0 | 0 | — |
case-07 | fail→fail | 36,089 | 36,956 | +2% | 1 | 1 | 0% | 4,805 | 5,924 | +23% | 0 | 0 | — |
case-08 | fail→fail | 39,590 | 53,323 | +35% | 1 | 1 | 0% | 6,359 | 8,718 | +37% | 0 | 0 | — |
case-09 | fail→fail | 22,370 | 56,043 | +151% | 1 | 1 | 0% | 3,299 | 8,808 | +167% | 0 | 0 | — |
case-10 | fail→pass | 55,340 | 49,780 | -10% | 1 | 1 | 0% | 8,241 | 8,809 | +7% | 0 | 0 | — |
case-11 | fail→fail | 37,772 | 48,669 | +29% | 1 | 1 | 0% | 5,588 | 8,237 | +47% | 0 | 0 | — |
case-12 | fail→fail | 39,736 | 52,940 | +33% | 1 | 1 | 0% | 5,993 | 8,817 | +47% | 0 | 0 | — |
case-13 | fail→fail | 42,409 | 52,041 | +23% | 1 | 1 | 0% | 6,746 | 8,818 | +31% | 0 | 0 | — |
case-14 | pass→pass | 25,026 | 42,740 | +71% | 1 | 1 | 0% | 3,728 | 6,843 | +84% | 0 | 0 | — |
case-15 | pass→pass | 33,928 | 47,938 | +41% | 1 | 1 | 0% | 5,109 | 7,850 | +54% | 0 | 0 | — |
case-16 | fail→pass | 36,612 | 54,531 | +49% | 1 | 1 | 0% | 5,988 | 8,804 | +47% | 0 | 0 | — |
case-17 | fail→fail | 49,003 | 46,353 | -5% | 1 | 1 | 0% | 7,122 | 7,171 | +1% | 0 | 0 | — |
case-18 | fail→pass | 39,310 | 52,216 | +33% | 1 | 1 | 0% | 6,155 | 8,813 | +43% | 0 | 0 | — |
case-19 | fail→fail | 25,116 | 12,561 | -50% | 1 | 1 | 0% | 3,769 | 1,340 | -64% | 0 | 0 | — |
case-20 | fail→fail | 21,939 | 44,571 | +103% | 1 | 1 | 0% | 3,345 | 7,471 | +123% | 0 | 0 | — |
case-21 | fail→fail | 67,276 | 10,411 | -85% | 1 | 1 | 0% | 2,368 | 1,134 | -52% | 0 | 0 | — |
case-22 | fail→fail | 51,540 | 52,926 | +3% | 1 | 1 | 0% | 7,790 | 8,818 | +13% | 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 20 counted toward the lift figure. The other 2 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 +5 percentage points is the difference between those two pass rates over the 20 comparable cases. 5 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.