Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Detect and validate emergent properties from combinations. Orchestrates emergent-property-identification → blend-elaboration.
.claude/skills/yogsoth-ai-emergence-detection/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 24% | 0% |
Detect and validate emergent properties from combinations — properties that exist in the combination but not in any individual component.
Use emergent-property-identification SOP to systematically scan combinations for non-additive properties. Compare predicted additive properties against actual combination properties.
Use blend-elaboration SOP to run the combination as a mental simulation, discovering additional emergent properties through dynamic interaction of combined elements.
| Metric | Floor | |--------|-------| | Emergent properties identified | ≥2 | | Properties verified as non-additive | ≥2 | | Emergence mechanism described | ≥1 per property | | Verification direction proposed | ≥1 per property |
| SOP | Role | |-----|------| | emergent-property-identification | Stage 1 — identify non-additive properties | | blend-elaboration | Stage 2 — simulate combination dynamics | | vital-relation-mapping | Supporting — map relations enabling emergence | | combinatorial-synthesis | Post — synthesize emergence findings |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→pass | 17,690 | 22,514 | +27% | 1 | 1 | 0% | 2,708 | 3,603 | +33% | 0 | 0 | — |
case-01 | pass→pass | 18,031 | 25,460 | +41% | 1 | 1 | 0% | 2,706 | 4,175 | +54% | 0 | 0 | — |
case-02 | pass→pass | 17,003 | 22,485 | +32% | 1 | 1 | 0% | 2,944 | 3,749 | +27% | 0 | 0 | — |
case-03 | pass→pass | 23,286 | 33,174 | +42% | 1 | 1 | 0% | 3,453 | 4,954 | +43% | 0 | 0 | — |
case-04 | pass→fail | 36,650 | 42,101 | +15% | 1 | 1 | 0% | 5,195 | 6,417 | +24% | 0 | 0 | — |
case-06 | pass→pass | 12,128 | 14,207 | +17% | 1 | 1 | 0% | 1,938 | 2,328 | +20% | 0 | 0 | — |
case-07 | fail→pass | 16,252 | 7,630 | -53% | 1 | 1 | 0% | 2,531 | 1,425 | -44% | 0 | 0 | — |
case-08 | pass→pass | 13,365 | 12,829 | -4% | 1 | 1 | 0% | 1,876 | 2,275 | +21% | 0 | 0 | — |
case-09 | pass→pass | 22,634 | 39,031 | +72% | 1 | 1 | 0% | 3,639 | 6,438 | +77% | 0 | 0 | — |
case-10 | pass→pass | 21,202 | 39,400 | +86% | 1 | 1 | 0% | 2,926 | 6,102 | +109% | 0 | 0 | — |
case-11 | fail→pass | 18,606 | 18,479 | -1% | 1 | 1 | 0% | 2,789 | 2,982 | +7% | 0 | 0 | — |
case-12 | pass→pass | 12,353 | 6,851 | -45% | 1 | 1 | 0% | 1,762 | 1,308 | -26% | 0 | 0 | — |
case-13 | pass→pass | 16,409 | 26,769 | +63% | 1 | 1 | 0% | 2,932 | 4,498 | +53% | 0 | 0 | — |
case-14 | pass→pass | 9,898 | 15,556 | +57% | 1 | 1 | 0% | 1,442 | 2,415 | +67% | 0 | 0 | — |
case-15 | pass→pass | 29,267 | 33,027 | +13% | 1 | 1 | 0% | 4,556 | 5,575 | +22% | 0 | 0 | — |
case-16 | fail→pass | 27,485 | 26,237 | -5% | 1 | 1 | 0% | 4,022 | 4,092 | +2% | 0 | 0 | — |
case-17 | pass→pass | 25,640 | 35,793 | +40% | 1 | 1 | 0% | 3,699 | 5,708 | +54% | 0 | 0 | — |
case-18 | pass→pass | 27,461 | 33,229 | +21% | 1 | 1 | 0% | 4,084 | 5,160 | +26% | 0 | 0 | — |
case-19 | fail→pass | 26,377 | 29,507 | +12% | 1 | 1 | 0% | 3,898 | 4,561 | +17% | 0 | 0 | — |
case-20 | pass→fail | 18,298 | 26,125 | +43% | 1 | 1 | 0% | 3,263 | 4,729 | +45% | 0 | 0 | — |
case-21 | pass→pass | 8,579 | 27,122 | +216% | 1 | 1 | 0% | 1,560 | 4,313 | +176% | 0 | 0 | — |
case-22 | pass→fail | 13,752 | 19,696 | +43% | 1 | 1 | 0% | 2,574 | 3,588 | +39% | 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 +5 percentage points is the difference between those two pass rates over the 22 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.