Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Propagate input uncertainties through the model via Monte Carlo sampling. Identifies which input uncertainties contribute most to output uncertainty.
.claude/skills/yogsoth-ai-uncertainty-propagation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 52% | 0% |
Map how input uncertainty flows to output uncertainty.
| Base SOP | Target | ±10% Range | |----------|--------|------------| | web-search | 20 | 18–22 | | web-research | 10 | 9–11 | | paper-overview | 25 | 22–28 | | paper-search | 20 | 18–22 | | paper-research | 10 | 9–11 |
<HARD-GATE>
| SOP | Done | Target | % |
|-----|------|--------|---|
| web-search | ? | 20 | ? |
| web-research | ? | 10 | ? |
| paper-overview | ? | 25 | ? |
| paper-search | ? | 20 | ? |
| paper-research | ? | 10 | ? |
Budget Gate: OPEN/CLOSED (>=80% required to exit)
</HARD-GATE>Import: web-search, web-research, paper-overview, paper-search, paper-research Subagent: distribution-assignment, monte-carlo-sampling, critical-path-identification
Assign probability distributions to uncertain inputs, propagate through the model via Monte Carlo sampling, analyze output distribution, identify which input uncertainties contribute most.
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | uncertainty-cascade | Uncertainty cascade propagation — assign input distributions, sample via Monte Carlo, propagate through model, analyze output distribution, identify critical paths. Maps how input uncertainty flows to output uncertainty. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | critical-path-identification | Identify which input uncertainties contribute most to output uncertainty and compute EVPI for research prioritization. | | distribution-assignment | Assign probability distributions to uncertain parameters based on available evidence and domain knowledge. | | monte-carlo-sampling | Design and execute Monte Carlo sampling strategy for uncertainty propagation through a model. |
<!-- 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→pass | 37,880 | 48,746 | +29% | 1 | 1 | 0% | 6,077 | 8,791 | +45% | 0 | 0 | — |
case-02 | fail→fail | 26,802 | 49,657 | +85% | 1 | 1 | 0% | 3,949 | 8,789 | +123% | 0 | 0 | — |
case-03 | fail→fail | 35,361 | 21,271 | -40% | 1 | 1 | 0% | 5,926 | 1,288 | -78% | 0 | 0 | — |
case-04 | fail→fail | 91,365 | 47,439 | -48% | 1 | 1 | 0% | 7,399 | 8,789 | +19% | 0 | 0 | — |
case-05 | fail→fail | 45,257 | 50,225 | +11% | 1 | 1 | 0% | 8,244 | 8,778 | +6% | 0 | 0 | — |
case-06 | fail→fail | 47,424 | 46,431 | -2% | 1 | 1 | 0% | 8,255 | 8,789 | +6% | 0 | 0 | — |
case-07 | fail→fail | 36,584 | 54,158 | +48% | 1 | 1 | 0% | 6,851 | 8,776 | +28% | 0 | 0 | — |
case-08 | fail→pass | 34,236 | 49,166 | +44% | 1 | 1 | 0% | 6,616 | 8,774 | +33% | 0 | 0 | — |
case-09 | fail→fail | 43,134 | 11,670 | -73% | 1 | 1 | 0% | 8,242 | 1,622 | -80% | 0 | 0 | — |
case-10 | fail→pass | 27,201 | 46,258 | +70% | 1 | 1 | 0% | 5,057 | 8,771 | +73% | 0 | 0 | — |
case-11 | fail→fail | 37,614 | 10,894 | -71% | 1 | 1 | 0% | 7,363 | 1,516 | -79% | 0 | 0 | — |
case-12 | fail→pass | 33,342 | 45,239 | +36% | 1 | 1 | 0% | 6,570 | 8,766 | +33% | 0 | 0 | — |
case-13 | fail→pass | 31,622 | 46,336 | +47% | 1 | 1 | 0% | 5,770 | 8,769 | +52% | 0 | 0 | — |
case-14 | fail→fail | 25,497 | 49,704 | +95% | 1 | 1 | 0% | 4,456 | 8,769 | +97% | 0 | 0 | — |
case-15 | fail→pass | 37,184 | 42,894 | +15% | 1 | 1 | 0% | 6,719 | 8,669 | +29% | 0 | 0 | — |
case-16 | fail→fail | 34,582 | 39,183 | +13% | 1 | 1 | 0% | 6,638 | 8,769 | +32% | 0 | 0 | — |
case-17 | fail→fail | 26,502 | 47,082 | +78% | 1 | 1 | 0% | 5,174 | 8,774 | +70% | 0 | 0 | — |
case-18 | fail→fail | 44,384 | 50,896 | +15% | 1 | 1 | 0% | 8,230 | 8,764 | +6% | 0 | 0 | — |
case-19 | fail→fail | 28,365 | 38,822 | +37% | 1 | 1 | 0% | 5,734 | 8,764 | +53% | 0 | 0 | — |
case-20 | pass→pass | 12,462 | 37,148 | +198% | 1 | 1 | 0% | 2,697 | 8,221 | +205% | 0 | 0 | — |
case-21 | pass→pass | 5,869 | 32,149 | +448% | 1 | 1 | 0% | 1,416 | 7,660 | +441% | 0 | 0 | — |
case-22 | pass→pass | 10,162 | 14,193 | +40% | 1 | 1 | 0% | 1,796 | 3,016 | +68% | 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 20 counted toward the lift figure. The other 2 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 20 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.