Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Remove components while preserving function via TRIZ trimming methodology. Simplify systems by redistributing functions.
.claude/skills/yogsoth-ai-function-trimming/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 82% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 48% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 12% | 0% |
| case-07 | ✓→✗ | ▼ Worse | -62% | 0% |
Systematically remove components while preserving or redistributing their functions to remaining elements.
| Resource | Target | Current | % | |----------|--------|---------|---| | web-search | 25 | 0 | 0% | | web-research | 8 | 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 | |--------|------| | component-decomposition | Identify components and trimming candidates | | combination-mapping | Map function redistribution options |
| SOP | Role | |-----|------| | function-model-construction | Build functional model with useful/harmful annotations | | trimming-execution | Progressive removal with function verification | | recombination-generation | Reassemble trimmed system | | structural-synthesis | Synthesize trimming outcomes |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | component-decomposition | Decompose system into functional components, identify dependencies, and surface trimming candidates. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | function-model-construction | Build substance-field functional model of a system, annotating useful, harmful, insufficient, and excessive interactions. | | structural-synthesis | Synthesize all structural transformation outputs into a coherent, ranked idea report with lineage tracking. | | trimming-execution | Progressively remove components from a system while verifying function preservation through redistribution. |
<!-- 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-01 | fail→fail | 19,041 | 32,842 | +72% | 1 | 1 | 0% | 2,989 | 4,763 | +59% | 0 | 0 | — |
case-02 | fail→fail | 22,527 | 28,016 | +24% | 1 | 1 | 0% | 3,666 | 5,328 | +45% | 0 | 0 | — |
case-03 | fail→pass | 23,077 | 34,632 | +50% | 1 | 1 | 0% | 3,711 | 6,761 | +82% | 0 | 0 | — |
case-04 | pass→fail | 18,172 | 23,875 | +31% | 1 | 1 | 0% | 3,052 | 4,523 | +48% | 0 | 0 | — |
case-05 | pass→pass | 11,675 | 13,953 | +20% | 1 | 1 | 0% | 2,700 | 3,541 | +31% | 0 | 0 | — |
case-06 | pass→fail | 25,588 | 28,086 | +10% | 1 | 1 | 0% | 6,000 | 6,724 | +12% | 0 | 0 | — |
case-07 | pass→fail | 26,101 | 6,079 | -77% | 1 | 1 | 0% | 4,199 | 1,581 | -62% | 0 | 0 | — |
case-08 | pass→fail | 26,269 | 9,102 | -65% | 1 | 1 | 0% | 4,267 | 1,188 | -72% | 0 | 0 | — |
case-09 | fail→fail | 24,417 | 33,447 | +37% | 1 | 1 | 0% | 4,207 | 6,370 | +51% | 0 | 0 | — |
case-10 | fail→pass | 26,908 | 27,225 | +1% | 1 | 1 | 0% | 4,325 | 4,990 | +15% | 0 | 0 | — |
case-11 | pass→pass | 24,426 | 36,192 | +48% | 1 | 1 | 0% | 4,318 | 6,134 | +42% | 0 | 0 | — |
case-12 | pass→fail | 26,134 | 7,477 | -71% | 1 | 1 | 0% | 4,755 | 1,068 | -78% | 0 | 0 | — |
case-13 | fail→fail | 20,645 | 8,277 | -60% | 1 | 1 | 0% | 3,404 | 1,045 | -69% | 0 | 0 | — |
case-14 | pass→pass | 22,593 | 33,926 | +50% | 1 | 1 | 0% | 3,992 | 6,491 | +63% | 0 | 0 | — |
case-15 | pass→fail | 24,677 | 10,442 | -58% | 1 | 1 | 0% | 4,187 | 1,246 | -70% | 0 | 0 | — |
case-16 | pass→fail | 21,930 | 8,354 | -62% | 1 | 1 | 0% | 3,725 | 1,149 | -69% | 0 | 0 | — |
case-17 | pass→pass | 23,157 | 16,461 | -29% | 1 | 1 | 0% | 3,787 | 3,311 | -13% | 0 | 0 | — |
case-18 | pass→pass | 22,712 | 22,842 | +1% | 1 | 1 | 0% | 3,707 | 4,441 | +20% | 0 | 0 | — |
case-19 | pass→fail | 22,950 | 12,176 | -47% | 1 | 1 | 0% | 3,827 | 1,531 | -60% | 0 | 0 | — |
case-20 | pass→pass | 20,916 | 29,869 | +43% | 1 | 1 | 0% | 3,473 | 5,665 | +63% | 0 | 0 | — |
case-21 | pass→pass | 22,787 | 21,196 | -7% | 1 | 1 | 0% | 4,002 | 3,897 | -3% | 0 | 0 | — |
case-22 | pass→pass | 25,904 | 26,974 | +4% | 1 | 1 | 0% | 4,356 | 5,305 | +22% | 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 16 counted toward the lift figure. The other 6 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 -27 percentage points is the difference between those two pass rates over the 16 comparable cases. 9 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.