Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Scan and select maximally diverse source domains. Ensures creative search covers genuinely unrelated fields with high transfer potential.
.claude/skills/yogsoth-ai-domain-divergence/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -40% | 0% |
Scan and select maximally diverse source domains to maximize creative potential.
Use domain-scanning SOP to identify candidate source domains. Cast a wide net across sciences, arts, engineering, biology, social systems, mathematics, and everyday life.
Inject randomness via random-paper-entry SOP. Select papers from unexpected fields to break domain fixation and discover overlooked source domains.
Use random-word-stimulus SOP to generate additional domain candidates that pure search would miss. Random words point to concrete domains (e.g., "coral" → marine biology → reef self-organization).
Assess the collected domains for genuine diversity:
| Metric | Floor | |--------|-------| | Domains scanned | ≥8 | | Unrelated domains identified | ≥3 | | Domains with transfer potential | ≥3 | | Discipline categories covered | ≥3 |
| SOP | Role | |-----|------| | domain-scanning | Stage 1 — systematic domain search | | random-paper-entry | Stage 2 — random paper as domain pointer | | random-word-stimulus | Stage 3 — random word as domain pointer | | abstraction-extraction | Stage 4 — verify transfer potential via abstraction | | analogy-quality-assessment | Stage 4 — assess structural similarity depth |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | abstraction-extraction | Extract abstract principles from concrete domain cases. Strips domain-specific details to reveal transferable mechanisms. | | analogy-quality-assessment | Assess analogy depth (surface/structural/systemic). Determines whether an analogy warrants transfer investment. | | random-paper-entry | Select random paper facet as creative stimulus. Uses genuine randomness in paper selection to break domain fixation. |
<!-- 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-03 | fail→pass | 21,120 | 17,070 | -19% | 1 | 1 | 0% | 3,191 | 3,019 | -5% | 0 | 0 | — |
case-01 | fail→fail | 24,801 | 24,241 | -2% | 1 | 1 | 0% | 3,519 | 4,097 | +16% | 0 | 0 | — |
case-02 | fail→pass | 30,168 | 23,613 | -22% | 1 | 1 | 0% | 4,215 | 4,016 | -5% | 0 | 0 | — |
case-04 | pass→pass | 26,838 | 2,218 | -92% | 1 | 1 | 0% | 1,977 | 884 | -55% | 0 | 0 | — |
case-05 | pass→pass | 14,819 | 11,256 | -24% | 1 | 1 | 0% | 2,278 | 2,523 | +11% | 0 | 0 | — |
case-06 | pass→pass | 14,597 | 7,387 | -49% | 1 | 1 | 0% | 2,204 | 1,639 | -26% | 0 | 0 | — |
case-07 | pass→pass | 17,208 | 10,956 | -36% | 1 | 1 | 0% | 2,497 | 2,010 | -20% | 0 | 0 | — |
case-08 | fail→pass | 13,401 | 7,686 | -43% | 1 | 1 | 0% | 2,010 | 1,857 | -8% | 0 | 0 | — |
case-09 | pass→pass | 17,562 | 14,672 | -16% | 1 | 1 | 0% | 2,540 | 2,600 | +2% | 0 | 0 | — |
case-10 | pass→pass | 14,702 | 12,013 | -18% | 1 | 1 | 0% | 2,168 | 2,245 | +4% | 0 | 0 | — |
case-11 | fail→pass | 21,894 | 1,958 | -91% | 1 | 1 | 0% | 1,270 | 829 | -35% | 0 | 0 | — |
case-12 | fail→pass | 22,559 | 1,823 | -92% | 1 | 1 | 0% | 1,336 | 800 | -40% | 0 | 0 | — |
case-13 | fail→pass | 9,380 | 2,363 | -75% | 1 | 1 | 0% | 1,599 | 938 | -41% | 0 | 0 | — |
case-14 | fail→pass | 16,940 | 13,103 | -23% | 1 | 1 | 0% | 2,652 | 2,754 | +4% | 0 | 0 | — |
case-15 | fail→pass | 16,273 | 12,817 | -21% | 1 | 1 | 0% | 2,495 | 2,447 | -2% | 0 | 0 | — |
case-16 | pass→pass | 8,031 | 1,620 | -80% | 1 | 1 | 0% | 1,162 | 785 | -32% | 0 | 0 | — |
case-17 | pass→pass | 14,396 | 6,993 | -51% | 1 | 1 | 0% | 2,150 | 1,558 | -28% | 0 | 0 | — |
case-18 | pass→pass | 11,686 | 2,193 | -81% | 1 | 1 | 0% | 1,656 | 876 | -47% | 0 | 0 | — |
case-19 | pass→pass | 15,412 | 9,156 | -41% | 1 | 1 | 0% | 2,491 | 1,937 | -22% | 0 | 0 | — |
case-20 | pass→pass | 17,751 | 18,864 | +6% | 1 | 1 | 0% | 3,552 | 4,198 | +18% | 0 | 0 | — |
case-21 | pass→pass | 15,894 | 14,038 | -12% | 1 | 1 | 0% | 2,214 | 2,678 | +21% | 0 | 0 | — |
case-22 | fail→pass | 21,794 | 24,622 | +13% | 1 | 1 | 0% | 3,327 | 4,150 | +25% | 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 +41 percentage points is the difference between those two pass rates over the 20 comparable cases.
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.