Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Construct complete 4-space blends with emergent structure. Orchestrates input-space-construction → generic-space-extraction → blend-composition.
.claude/skills/yogsoth-ai-blend-construction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 84% | 0% |
Construct complete 4-space blends with emergent structure following the Fauconnier-Turner conceptual integration network model.
Build rich input spaces for both source concepts using input-space-construction SOP. Each space must include elements, relations, attributes, and internal logic.
Extract the shared abstract structure from both input spaces using generic-space-extraction SOP. The generic space captures what the two inputs have in common at the most abstract level.
Compose the blended space by selectively projecting structure from both inputs and creating new connections using blend-composition SOP. The blend must develop emergent structure not present in either input.
| Metric | Floor | |--------|-------| | Complete 4-space blends | ≥2 | | Emergent structures per blend | ≥1 | | Vital relations compressed | ≥3 per blend | | Novel connections in blend | ≥2 per blend |
| SOP | Role | |-----|------| | input-space-construction | Stage 1 — build input spaces | | generic-space-extraction | Stage 2 — extract shared structure | | blend-composition | Stage 3 — compose blended space | | blend-completion | Post-Stage 3 — recruit background knowledge | | vital-relation-mapping | Pre-Stage 1 — map vital relations to guide projection |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 30,400 | 34,125 | +12% | 1 | 1 | 0% | 4,845 | 5,810 | +20% | 0 | 0 | — |
case-02 | fail→pass | 26,662 | 30,665 | +15% | 1 | 1 | 0% | 3,780 | 5,080 | +34% | 0 | 0 | — |
case-03 | fail→pass | 26,395 | 37,581 | +42% | 1 | 1 | 0% | 4,113 | 6,496 | +58% | 0 | 0 | — |
case-04 | pass→pass | 12,492 | 10,149 | -19% | 1 | 1 | 0% | 2,117 | 2,122 | +0% | 0 | 0 | — |
case-05 | pass→pass | 21,963 | 26,377 | +20% | 1 | 1 | 0% | 3,887 | 5,038 | +30% | 0 | 0 | — |
case-06 | pass→pass | 21,364 | 34,679 | +62% | 1 | 1 | 0% | 3,454 | 6,492 | +88% | 0 | 0 | — |
case-07 | pass→pass | 17,296 | 30,642 | +77% | 1 | 1 | 0% | 2,960 | 5,319 | +80% | 0 | 0 | — |
case-08 | pass→pass | 18,879 | 35,415 | +88% | 1 | 1 | 0% | 2,836 | 6,052 | +113% | 0 | 0 | — |
case-09 | pass→pass | 24,572 | 32,228 | +31% | 1 | 1 | 0% | 3,997 | 5,200 | +30% | 0 | 0 | — |
case-10 | pass→pass | 22,215 | 39,126 | +76% | 1 | 1 | 0% | 3,391 | 6,437 | +90% | 0 | 0 | — |
case-11 | pass→pass | 21,516 | 30,446 | +42% | 1 | 1 | 0% | 3,486 | 5,092 | +46% | 0 | 0 | — |
case-12 | pass→pass | 35,006 | 37,502 | +7% | 1 | 1 | 0% | 5,471 | 6,199 | +13% | 0 | 0 | — |
case-13 | fail→pass | 18,962 | 29,734 | +57% | 1 | 1 | 0% | 3,024 | 4,954 | +64% | 0 | 0 | — |
case-14 | pass→pass | 20,957 | 38,280 | +83% | 1 | 1 | 0% | 3,642 | 6,499 | +78% | 0 | 0 | — |
case-15 | pass→pass | 20,129 | 36,378 | +81% | 1 | 1 | 0% | 3,082 | 6,115 | +98% | 0 | 0 | — |
case-16 | pass→pass | 31,477 | 33,148 | +5% | 1 | 1 | 0% | 4,757 | 5,760 | +21% | 0 | 0 | — |
case-17 | pass→pass | 19,090 | 32,098 | +68% | 1 | 1 | 0% | 2,884 | 5,161 | +79% | 0 | 0 | — |
case-18 | pass→pass | 29,198 | 40,653 | +39% | 1 | 1 | 0% | 4,313 | 6,496 | +51% | 0 | 0 | — |
case-19 | pass→pass | 19,307 | 33,618 | +74% | 1 | 1 | 0% | 2,976 | 5,288 | +78% | 0 | 0 | — |
case-20 | fail→pass | 22,682 | 40,539 | +79% | 1 | 1 | 0% | 3,409 | 6,287 | +84% | 0 | 0 | — |
case-21 | pass→pass | 20,023 | 37,633 | +88% | 1 | 1 | 0% | 2,772 | 5,554 | +100% | 0 | 0 | — |
case-22 | fail→pass | 41,147 | 33,541 | -18% | 1 | 1 | 0% | 5,752 | 5,555 | -3% | 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 +27 percentage points is the difference between those two pass rates over the 22 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.