Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Challenge industry best practices' hidden assumptions. Deconstruct benchmarks to reveal unexamined constraints.
.claude/skills/yogsoth-ai-anti-benchmark/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -32% | 0% |
Challenge industry best practices' hidden assumptions.
| Resource | Target | Current | % | |----------|--------|---------|---| | web-search | 25 | 0 | 0% | | web-research | 10 | 0 | 0% | | paper-overview | 25 | 0 | 0% | | paper-search | 15 | 0 | 0% | | paper-research | 5 | 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 | |--------|------| | assumption-enumeration | Surface assumptions hidden in benchmarks |
| SOP | Role | |-----|------| | benchmark-challenge | Identify and negate benchmark assumptions | | sacred-cow-identification | Find unquestioned beliefs behind best practices | | assumption-perturbation | Test what happens when benchmark assumptions fail | | constructive-rebellion | Build alternatives that violate benchmarks constructively | | destruction-synthesis | Synthesize anti-benchmark outputs |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | creative-ideation-assumption-enumeration | Surface, perturb, and prioritize assumptions by disruption potential. Orchestrates assumption surfacing → perturbation → sacred cow identification → prioritization. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | benchmark-challenge | Identify and negate benchmark assumptions. Deconstruct best practices to reveal hidden constraints and open new spaces. | | constructive-rebellion | Build constructive alternatives from destructive negation. Transform violated assumptions into viable innovation directions. | | creative-ideation-assumption-perturbation | Perturb each assumption, observe system response. Systematic stress-testing of assumptions to reveal fragility and opportunity. | | destruction-synthesis | Synthesize all assumption destruction outputs into structured destructive innovation report. | | sacred-cow-identification | Find domain's unquestioned beliefs. Systematic identification of dogma that constrains innovation. |
<!-- 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 | pass→pass | 13,084 | 2,596 | -80% | 1 | 1 | 0% | 2,056 | 1,016 | -51% | 0 | 0 | — |
case-02 | fail→pass | 16,319 | 1,840 | -89% | 1 | 1 | 0% | 815 | 972 | +19% | 0 | 0 | — |
case-03 | fail→pass | 13,586 | 2,319 | -83% | 1 | 1 | 0% | 654 | 1,031 | +58% | 0 | 0 | — |
case-04 | fail→pass | 7,983 | 2,024 | -75% | 1 | 1 | 0% | 1,255 | 943 | -25% | 0 | 0 | — |
case-05 | fail→pass | 10,208 | 2,240 | -78% | 1 | 1 | 0% | 1,548 | 1,067 | -31% | 0 | 0 | — |
case-06 | fail→pass | 9,324 | 1,744 | -81% | 1 | 1 | 0% | 1,375 | 940 | -32% | 0 | 0 | — |
case-07 | fail→pass | 8,450 | 4,900 | -42% | 1 | 1 | 0% | 1,332 | 1,080 | -19% | 0 | 0 | — |
case-12 | fail→pass | 11,480 | 1,975 | -83% | 1 | 1 | 0% | 1,716 | 1,061 | -38% | 0 | 0 | — |
case-08 | pass→pass | 9,426 | 7,334 | -22% | 1 | 1 | 0% | 1,495 | 1,923 | +29% | 0 | 0 | — |
case-09 | fail→pass | 18,049 | 3,869 | -79% | 1 | 1 | 0% | 2,925 | 1,282 | -56% | 0 | 0 | — |
case-10 | fail→pass | 11,468 | 3,832 | -67% | 1 | 1 | 0% | 1,715 | 1,265 | -26% | 0 | 0 | — |
case-11 | pass→pass | 15,290 | 3,743 | -76% | 1 | 1 | 0% | 2,283 | 1,295 | -43% | 0 | 0 | — |
case-13 | fail→pass | 14,031 | 2,293 | -84% | 1 | 1 | 0% | 2,183 | 1,070 | -51% | 0 | 0 | — |
case-14 | pass→fail | 14,296 | 14,319 | +0% | 1 | 1 | 0% | 2,397 | 1,885 | -21% | 0 | 0 | — |
case-15 | pass→pass | 13,957 | 3,192 | -77% | 1 | 1 | 0% | 2,141 | 1,147 | -46% | 0 | 0 | — |
case-16 | fail→pass | 13,303 | 3,876 | -71% | 1 | 1 | 0% | 2,117 | 1,364 | -36% | 0 | 0 | — |
case-17 | pass→fail | 16,737 | 9,256 | -45% | 1 | 1 | 0% | 2,494 | 1,302 | -48% | 0 | 0 | — |
case-18 | pass→pass | 16,880 | 2,558 | -85% | 1 | 1 | 0% | 2,815 | 1,075 | -62% | 0 | 0 | — |
case-19 | fail→pass | 13,274 | 3,079 | -77% | 1 | 1 | 0% | 2,100 | 1,160 | -45% | 0 | 0 | — |
case-20 | pass→fail | 14,484 | 30,316 | +109% | 1 | 1 | 0% | 2,873 | 5,747 | +100% | 0 | 0 | — |
case-21 | pass→fail | 16,442 | 32,459 | +97% | 1 | 1 | 0% | 2,767 | 5,619 | +103% | 0 | 0 | — |
case-22 | pass→fail | 16,394 | 35,624 | +117% | 1 | 1 | 0% | 2,847 | 6,412 | +125% | 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 18 counted toward the lift figure. The other 4 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 +32 percentage points is the difference between those two pass rates over the 18 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.