Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Execute component surgery operations (subtract/multiply/divide/unify/redirect) from Systematic Inventive Thinking.
.claude/skills/yogsoth-ai-surgery-operation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-20 | ✓→✗ | ▼ Worse | -87% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | 26% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 8% | 0% |
| case-02 | ✗→✗ | ＝ Same ✗ | -18% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | 31% | 0% |
Apply SIT surgical operators to system components for structural innovation.
Subagent — spawned via subagent-spawning/spawn-agent skill.
Component surgery requires careful reasoning about system integrity during each operation. Benefits from dedicated context that can track component states and verify function preservation after each cut.
<!-- 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-02 | fail→fail | 44,034 | 35,526 | -19% | 1 | 1 | 0% | 6,060 | 4,991 | -18% | 0 | 0 | — |
case-01 | fail→fail | 20,427 | 31,044 | +52% | 1 | 1 | 0% | 3,204 | 4,203 | +31% | 0 | 0 | — |
case-03 | fail→fail | 42,564 | 37,149 | -13% | 1 | 1 | 0% | 7,287 | 5,640 | -23% | 0 | 0 | — |
case-04 | fail→fail | 17,313 | 22,966 | +33% | 1 | 1 | 0% | 2,787 | 3,872 | +39% | 0 | 0 | — |
case-05 | fail→fail | 22,630 | 21,590 | -5% | 1 | 1 | 0% | 3,474 | 2,604 | -25% | 0 | 0 | — |
case-06 | fail→fail | 15,531 | 21,253 | +37% | 1 | 1 | 0% | 2,540 | 2,581 | +2% | 0 | 0 | — |
case-07 | fail→fail | 21,701 | 31,417 | +45% | 1 | 1 | 0% | 3,655 | 5,043 | +38% | 0 | 0 | — |
case-08 | fail→fail | 35,357 | 55,776 | +58% | 1 | 1 | 0% | 5,910 | 6,904 | +17% | 0 | 0 | — |
case-09 | fail→fail | 19,059 | 9,877 | -48% | 1 | 1 | 0% | 2,400 | 556 | -77% | 0 | 0 | — |
case-10 | fail→fail | 23,202 | 26,598 | +15% | 1 | 1 | 0% | 3,173 | 3,502 | +10% | 0 | 0 | — |
case-11 | fail→fail | 19,711 | 32,875 | +67% | 1 | 1 | 0% | 3,342 | 3,048 | -9% | 0 | 0 | — |
case-12 | fail→fail | 62,548 | 40,152 | -36% | 1 | 1 | 0% | 5,033 | 4,702 | -7% | 0 | 0 | — |
case-13 | fail→fail | 33,679 | 33,373 | -1% | 1 | 1 | 0% | 4,987 | 5,479 | +10% | 0 | 0 | — |
case-14 | fail→fail | 17,679 | 26,431 | +50% | 1 | 1 | 0% | 3,426 | 730 | -79% | 0 | 0 | — |
case-15 | fail→fail | 25,193 | 33,661 | +34% | 1 | 1 | 0% | 3,823 | 5,096 | +33% | 0 | 0 | — |
case-16 | fail→fail | 25,380 | 32,818 | +29% | 1 | 1 | 0% | 4,416 | 4,534 | +3% | 0 | 0 | — |
case-17 | fail→fail | 27,487 | 26,978 | -2% | 1 | 1 | 0% | 3,301 | 3,845 | +16% | 0 | 0 | — |
case-18 | fail→fail | 19,858 | 40,738 | +105% | 1 | 1 | 0% | 4,070 | 6,886 | +69% | 0 | 0 | — |
case-19 | fail→fail | 19,026 | 11,623 | -39% | 1 | 1 | 0% | 3,068 | 411 | -87% | 0 | 0 | — |
case-20 | pass→fail | 13,770 | 18,708 | +36% | 1 | 1 | 0% | 2,604 | 343 | -87% | 0 | 0 | — |
case-21 | pass→pass | 17,189 | 18,010 | +5% | 1 | 1 | 0% | 1,966 | 2,476 | +26% | 0 | 0 | — |
case-22 | pass→pass | 12,048 | 11,122 | -8% | 1 | 1 | 0% | 1,166 | 1,265 | +8% | 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 18 counted toward the lift figure. The other 4 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 18 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.