Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Rate failure mode occurrence probability 1-10. Estimates how likely each failure mode is to manifest during research execution.
.claude/skills/yogsoth-ai-occurrence-scoring/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -68% | 0% |
Rates each failure mode's occurrence probability on a 1-10 scale.
Subagent — spawned via subagent-spawning/spawn-agent.
Occurrence estimation requires probabilistic reasoning isolated from severity and detection considerations. Prevents halo effects.
<!-- 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-07 | fail→pass | 34,425 | 17,872 | -48% | 1 | 1 | 0% | 2,754 | 1,956 | -29% | 0 | 0 | — |
case-06 | pass→pass | 30,079 | 13,239 | -56% | 1 | 1 | 0% | 1,288 | 515 | -60% | 0 | 0 | — |
case-01 | pass→pass | 48,020 | 33,862 | -29% | 1 | 1 | 0% | 3,404 | 2,608 | -23% | 0 | 0 | — |
case-02 | fail→pass | 28,179 | 11,657 | -59% | 1 | 1 | 0% | 2,376 | 1,001 | -58% | 0 | 0 | — |
case-03 | fail→pass | 26,023 | 7,497 | -71% | 1 | 1 | 0% | 2,056 | 596 | -71% | 0 | 0 | — |
case-04 | fail→pass | 16,161 | 11,031 | -32% | 1 | 1 | 0% | 1,391 | 633 | -54% | 0 | 0 | — |
case-05 | pass→pass | 34,042 | 26,203 | -23% | 1 | 1 | 0% | 2,883 | 779 | -73% | 0 | 0 | — |
case-08 | pass→pass | 22,895 | 20,286 | -11% | 1 | 1 | 0% | 2,754 | 2,161 | -22% | 0 | 0 | — |
case-09 | pass→pass | 34,669 | 19,053 | -45% | 1 | 1 | 0% | 2,106 | 1,962 | -7% | 0 | 0 | — |
case-10 | fail→pass | 45,175 | 22,042 | -51% | 1 | 1 | 0% | 1,549 | 500 | -68% | 0 | 0 | — |
case-11 | pass→pass | 14,792 | 25,375 | +72% | 1 | 1 | 0% | 1,579 | 526 | -67% | 0 | 0 | — |
case-12 | fail→fail | 14,387 | 9,240 | -36% | 1 | 1 | 0% | 1,651 | 469 | -72% | 0 | 0 | — |
case-13 | pass→pass | 19,202 | 8,299 | -57% | 1 | 1 | 0% | 1,492 | 749 | -50% | 0 | 0 | — |
case-14 | fail→pass | 39,141 | 2,285 | -94% | 1 | 1 | 0% | 1,604 | 537 | -67% | 0 | 0 | — |
case-15 | fail→pass | 15,529 | 14,148 | -9% | 1 | 1 | 0% | 2,205 | 2,143 | -3% | 0 | 0 | — |
case-16 | pass→pass | 11,652 | 8,032 | -31% | 1 | 1 | 0% | 2,032 | 596 | -71% | 0 | 0 | — |
case-17 | fail→pass | 18,966 | 14,116 | -26% | 1 | 1 | 0% | 2,295 | 572 | -75% | 0 | 0 | — |
case-18 | fail→pass | 20,172 | 10,729 | -47% | 1 | 1 | 0% | 2,715 | 1,046 | -61% | 0 | 0 | — |
case-19 | fail→pass | 15,194 | 10,269 | -32% | 1 | 1 | 0% | 2,147 | 1,049 | -51% | 0 | 0 | — |
case-20 | pass→pass | 12,271 | 7,068 | -42% | 1 | 1 | 0% | 852 | 499 | -41% | 0 | 0 | — |
case-21 | pass→pass | 15,099 | 8,778 | -42% | 1 | 1 | 0% | 1,347 | 514 | -62% | 0 | 0 | — |
case-22 | pass→pass | 19,471 | 17,091 | -12% | 1 | 1 | 0% | 2,214 | 2,060 | -7% | 0 | 0 | — |
case-23 | pass→pass | 16,929 | 23,174 | +37% | 1 | 1 | 0% | 2,841 | 3,071 | +8% | 0 | 0 | — |
case-24 | pass→pass | 16,723 | 12,008 | -28% | 1 | 1 | 0% | 1,616 | 1,343 | -17% | 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. 24 cases were attempted, and 23 counted toward the lift figure. The other 1 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 +42 percentage points is the difference between those two pass rates over the 23 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.