Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Synthesize all structural transformation outputs into a coherent, ranked idea report with lineage tracking.
.claude/skills/yogsoth-ai-structural-synthesis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | 106% | 0% |
| case-19 | ✓→✗ | ▼ Worse | 70% | 0% |
| case-22 | ✓→✗ | ▼ Worse | 46% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | -1% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -44% | 0% |
Synthesize all structural transformation outputs into a final report.
Subagent — spawned via subagent-spawning/spawn-agent skill.
Synthesis requires integrating outputs from multiple diverse operations, deduplicating, clustering, and ranking. Benefits from dedicated context that can hold all intermediate results simultaneously for cross-referencing.
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | spawn-agent | Spawn a customized CC subagent with full MCP tool access. Used by SOPs that declare execution: subagent. |
<!-- 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 | 12,226 | 23,190 | +90% | 1 | 1 | 0% | 1,047 | 591 | -44% | 0 | 0 | — |
case-02 | fail→fail | 12,203 | 21,537 | +76% | 1 | 1 | 0% | 1,134 | 1,084 | -4% | 0 | 0 | — |
case-03 | fail→fail | 8,946 | 33,632 | +276% | 1 | 1 | 0% | 552 | 4,935 | +794% | 0 | 0 | — |
case-04 | fail→fail | 33,730 | 19,449 | -42% | 1 | 1 | 0% | 4,886 | 554 | -89% | 0 | 0 | — |
case-05 | fail→fail | 23,816 | 20,038 | -16% | 1 | 1 | 0% | 2,969 | 740 | -75% | 0 | 0 | — |
case-06 | fail→fail | 14,631 | 17,342 | +19% | 1 | 1 | 0% | 1,380 | 719 | -48% | 0 | 0 | — |
case-07 | fail→fail | 49,153 | 15,940 | -68% | 1 | 1 | 0% | 8,226 | 492 | -94% | 0 | 0 | — |
case-08 | fail→fail | 15,518 | 19,962 | +29% | 1 | 1 | 0% | 1,649 | 753 | -54% | 0 | 0 | — |
case-09 | fail→fail | 41,415 | 17,048 | -59% | 1 | 1 | 0% | 6,385 | 416 | -93% | 0 | 0 | — |
case-10 | fail→fail | 33,860 | 14,621 | -57% | 1 | 1 | 0% | 4,740 | 414 | -91% | 0 | 0 | — |
case-11 | fail→fail | 24,416 | 16,625 | -32% | 1 | 1 | 0% | 3,252 | 480 | -85% | 0 | 0 | — |
case-12 | fail→fail | 23,441 | 45,782 | +95% | 1 | 1 | 0% | 2,937 | 6,956 | +137% | 0 | 0 | — |
case-13 | fail→fail | 25,830 | 18,446 | -29% | 1 | 1 | 0% | 3,585 | 492 | -86% | 0 | 0 | — |
case-14 | fail→fail | 23,927 | 17,997 | -25% | 1 | 1 | 0% | 3,667 | 551 | -85% | 0 | 0 | — |
case-15 | fail→fail | 39,482 | 14,212 | -64% | 1 | 1 | 0% | 6,225 | 492 | -92% | 0 | 0 | — |
case-16 | fail→fail | 27,919 | 15,512 | -44% | 1 | 1 | 0% | 4,357 | 506 | -88% | 0 | 0 | — |
case-17 | fail→fail | 38,743 | 14,061 | -64% | 1 | 1 | 0% | 5,709 | 383 | -93% | 0 | 0 | — |
case-18 | fail→fail | 30,911 | 37,829 | +22% | 1 | 1 | 0% | 4,808 | 6,517 | +36% | 0 | 0 | — |
case-19 | pass→fail | 8,310 | 32,622 | +293% | 1 | 1 | 0% | 532 | 906 | +70% | 0 | 0 | — |
case-20 | pass→pass | 20,225 | 18,581 | -8% | 1 | 1 | 0% | 2,683 | 2,666 | -1% | 0 | 0 | — |
case-21 | fail→pass | 8,944 | 12,975 | +45% | 1 | 1 | 0% | 709 | 1,461 | +106% | 0 | 0 | — |
case-22 | pass→fail | 7,790 | 9,093 | +17% | 1 | 1 | 0% | 527 | 769 | +46% | 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 6 counted toward the lift figure. The other 16 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 -5 percentage points is the difference between those two pass rates over the 6 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.