Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Extract transferable structural principles from source domains. Orchestrates source identification → abstraction → structural mapping → transfer validation.
.claude/skills/yogsoth-ai-analogy-extraction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 21% | 0% |
Extract transferable structural principles from source domains.
Identify candidate source domains using domain-scanning SOP. Evaluate each for structural similarity depth (surface/structural/systemic).
For each promising source, extract the abstract principle using abstraction-extraction or biological-strategy-extraction SOP. Strip domain-specific details to reveal the transferable mechanism.
Map source structure to target domain. Identify: corresponding elements, missing elements (gaps), extra elements (opportunities). Use structural-mapping SOP.
Assess mapping quality: Is the analogy surface-level (shared labels) or deep (shared relational structure)? Use analogy-quality-assessment SOP. Only deep analogies warrant transfer.
| Metric | Floor | |--------|-------| | Source domains scanned | ≥5 | | Abstractions extracted | ≥3 | | Structural mappings completed | ≥3 | | Validated deep analogies | ≥2 |
| SOP | Role | |-----|------| | domain-scanning | Stage 1 — find candidate source domains | | web-search | Stage 1 — supplement domain search | | paper-overview | Stage 1 — find academic analogies | | abstraction-extraction | Stage 2 — extract abstract principles | | structural-mapping | Stage 3 — map source→target structure | | analogy-quality-assessment | Stage 4 — validate mapping depth | | novelty-scoring | Post — score resulting ideas | | idea-synthesis | Post — synthesize into coherent concepts |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | pass→pass | 27,900 | 34,546 | +24% | 1 | 1 | 0% | 3,990 | 5,302 | +33% | 0 | 0 | — |
case-01 | fail→pass | 32,065 | 30,407 | -5% | 1 | 1 | 0% | 4,325 | 4,508 | +4% | 0 | 0 | — |
case-02 | fail→pass | 43,419 | 32,596 | -25% | 1 | 1 | 0% | 6,233 | 4,791 | -23% | 0 | 0 | — |
case-03 | fail→pass | 44,526 | 36,412 | -18% | 1 | 1 | 0% | 6,226 | 6,152 | -1% | 0 | 0 | — |
case-04 | pass→pass | 44,107 | 43,484 | -1% | 1 | 1 | 0% | 6,208 | 6,566 | +6% | 0 | 0 | — |
case-05 | fail→pass | 30,611 | 37,322 | +22% | 1 | 1 | 0% | 4,607 | 5,739 | +25% | 0 | 0 | — |
case-06 | fail→pass | 34,588 | 41,203 | +19% | 1 | 1 | 0% | 5,391 | 6,544 | +21% | 0 | 0 | — |
case-07 | pass→pass | 28,539 | 42,568 | +49% | 1 | 1 | 0% | 4,143 | 6,561 | +58% | 0 | 0 | — |
case-08 | fail→pass | 32,849 | 18,148 | -45% | 1 | 1 | 0% | 4,995 | 3,091 | -38% | 0 | 0 | — |
case-09 | pass→pass | 21,523 | 21,357 | -1% | 1 | 1 | 0% | 3,006 | 3,299 | +10% | 0 | 0 | — |
case-10 | fail→pass | 24,773 | 34,342 | +39% | 1 | 1 | 0% | 3,388 | 5,476 | +62% | 0 | 0 | — |
case-11 | pass→pass | 26,415 | 16,733 | -37% | 1 | 1 | 0% | 3,587 | 2,858 | -20% | 0 | 0 | — |
case-12 | pass→pass | 21,164 | 25,835 | +22% | 1 | 1 | 0% | 2,854 | 4,318 | +51% | 0 | 0 | — |
case-13 | fail→pass | 24,509 | 22,966 | -6% | 1 | 1 | 0% | 3,741 | 3,718 | -1% | 0 | 0 | — |
case-14 | pass→pass | 23,589 | 34,275 | +45% | 1 | 1 | 0% | 3,452 | 5,463 | +58% | 0 | 0 | — |
case-15 | fail→pass | 28,818 | 31,039 | +8% | 1 | 1 | 0% | 4,304 | 4,852 | +13% | 0 | 0 | — |
case-16 | pass→pass | 24,639 | 28,166 | +14% | 1 | 1 | 0% | 3,614 | 4,511 | +25% | 0 | 0 | — |
case-17 | fail→pass | 34,881 | 33,895 | -3% | 1 | 1 | 0% | 4,692 | 5,533 | +18% | 0 | 0 | — |
case-18 | pass→pass | 21,792 | 26,518 | +22% | 1 | 1 | 0% | 3,235 | 4,320 | +34% | 0 | 0 | — |
case-20 | pass→fail | 18,584 | 25,634 | +38% | 1 | 1 | 0% | 3,148 | 4,452 | +41% | 0 | 0 | — |
case-21 | pass→fail | 11,834 | 12,844 | +9% | 1 | 1 | 0% | 1,781 | 2,332 | +31% | 0 | 0 | — |
case-22 | pass→pass | 15,398 | 19,283 | +25% | 1 | 1 | 0% | 2,709 | 3,853 | +42% | 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 +36 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.