Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Systematic one-factor-at-a-time parameter sweep
.claude/skills/yogsoth-ai-parameter-variation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 95% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 52% | 0% |
Systematic one-factor-at-a-time (OFAT) parameter sweep to explore how individual dimension changes affect the solution space.
| Resource | Target | Current | % | |----------|--------|---------|---| | web-search | 20 | 0 | 0% | | web-research | 5 | 0 | 0% | | paper-overview | 20 | 0 | 0% | | paper-search | 12 | 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 | |--------|------| | combination-mapping | Map parameter-value combinations |
| SOP | Role | |-----|------| | value-enumeration | Enumerate values for sweep | | combination-evaluation | Evaluate each variation | | path-generation | Generate sweep paths | | morphological-synthesis | Synthesize variation report |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | morphological-synthesis | Synthesize all morphological exploration outputs | | path-generation | Generate combination paths through consistent space | | value-enumeration | Enumerate 3-5 values per parameter including extremes |
<!-- 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-02 | fail→pass | 24,784 | 33,102 | +34% | 1 | 1 | 0% | 2,517 | 4,899 | +95% | 0 | 0 | — |
case-01 | fail→fail | 47,987 | 50,272 | +5% | 1 | 1 | 0% | 7,946 | 8,688 | +9% | 0 | 0 | — |
case-03 | fail→pass | 22,316 | 32,193 | +44% | 1 | 1 | 0% | 2,610 | 4,973 | +91% | 0 | 0 | — |
case-04 | fail→pass | 17,542 | 16,217 | -8% | 1 | 1 | 0% | 2,149 | 2,419 | +13% | 0 | 0 | — |
case-05 | fail→pass | 9,666 | 8,963 | -7% | 1 | 1 | 0% | 1,424 | 1,067 | -25% | 0 | 0 | — |
case-06 | fail→fail | 15,361 | 43,227 | +181% | 1 | 1 | 0% | 2,168 | 6,376 | +194% | 0 | 0 | — |
case-07 | fail→fail | 15,893 | 21,553 | +36% | 1 | 1 | 0% | 1,563 | 3,102 | +98% | 0 | 0 | — |
case-08 | fail→fail | 42,313 | 6,014 | -86% | 1 | 1 | 0% | 2,648 | 1,405 | -47% | 0 | 0 | — |
case-09 | fail→fail | 21,819 | 7,532 | -65% | 1 | 1 | 0% | 2,763 | 833 | -70% | 0 | 0 | — |
case-10 | pass→pass | 18,046 | 4,064 | -77% | 1 | 1 | 0% | 2,002 | 1,079 | -46% | 0 | 0 | — |
case-11 | fail→pass | 25,549 | 31,715 | +24% | 1 | 1 | 0% | 3,316 | 5,047 | +52% | 0 | 0 | — |
case-12 | pass→pass | 18,369 | 25,754 | +40% | 1 | 1 | 0% | 2,776 | 3,700 | +33% | 0 | 0 | — |
case-13 | pass→fail | 17,655 | 19,724 | +12% | 1 | 1 | 0% | 1,899 | 1,149 | -39% | 0 | 0 | — |
case-14 | fail→fail | 13,790 | 8,878 | -36% | 1 | 1 | 0% | 1,271 | 964 | -24% | 0 | 0 | — |
case-15 | fail→fail | 16,513 | 24,197 | +47% | 1 | 1 | 0% | 1,601 | 3,171 | +98% | 0 | 0 | — |
case-16 | fail→pass | 13,126 | 5,811 | -56% | 1 | 1 | 0% | 1,145 | 1,284 | +12% | 0 | 0 | — |
case-17 | fail→fail | 9,707 | 8,384 | -14% | 1 | 1 | 0% | 1,434 | 944 | -34% | 0 | 0 | — |
case-18 | fail→fail | 12,225 | 8,877 | -27% | 1 | 1 | 0% | 1,356 | 1,036 | -24% | 0 | 0 | — |
case-19 | pass→pass | 16,826 | 20,579 | +22% | 1 | 1 | 0% | 2,125 | 3,624 | +71% | 0 | 0 | — |
case-20 | pass→pass | 18,717 | 38,536 | +106% | 1 | 1 | 0% | 2,711 | 6,838 | +152% | 0 | 0 | — |
case-21 | pass→fail | 33,340 | 42,940 | +29% | 1 | 1 | 0% | 5,425 | 8,675 | +60% | 0 | 0 | — |
case-22 | pass→fail | 24,754 | 57,375 | +132% | 1 | 1 | 0% | 4,929 | 8,682 | +76% | 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 21 counted toward the lift figure. The other 1 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 21 comparable cases. 3 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.