Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strategy: Probability of Necessity and Sufficiency (PNS/PS) — systematically evaluate whether each factor is necessary, sufficient, both, or neither for the conclusion.
.claude/skills/yogsoth-ai-necessity-sufficiency/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 98% | 0% |
PNS/PS framework: classify each factor by its causal role — necessary, sufficient, both, or neither.
| Parameter | S | M | L | |---|---|---|---| | Factors tested | 5 | 10 | 20 | | Necessity evaluations | 3 | 6 | 12 | | Sufficiency evaluations | 3 | 6 | 12 |
causal-claim-extraction → factor-enumeration
→ [for each factor]:
necessity-evaluation (PN score)
+ sufficiency-evaluation (PS score)
+ single-factor-removal (supporting evidence)
→ load-bearing-identification (quadrant classification)<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | causal-necessity-testing | Tactic: Extract causal claims, evaluate probability of necessity (PN) and sufficiency (PS) for each, classify into necessity-sufficiency quadrants. | | systematic-factor-ablation | Tactic: List all factors, remove one at a time, assess conclusion stability, rank factors by load-bearing importance. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | causal-claim-extraction | Extract all causal claims (X causes Y, X leads to Y, X enables Y) from an artifact, producing a structured list of cause-effect pairs. | | factor-enumeration | List all key factors, conditions, and assumptions that support or enable the artifact's conclusion. | | load-bearing-identification | Identify which factors are "load-bearing walls" — factors whose removal would collapse the conclusion. | | necessity-evaluation | Evaluate the probability of necessity (PN) for a causal factor — would the conclusion fail if this factor were absent? | | single-factor-removal | Remove one specified factor from the artifact's support structure and reason about how the conclusion changes. | | sufficiency-evaluation | Evaluate the probability of sufficiency (PS) for a causal factor — would this factor alone be enough to produce the conclusion? |
<!-- 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 | pass→pass | 29,056 | 16,738 | -42% | 1 | 1 | 0% | 1,821 | 2,797 | +54% | 0 | 0 | — |
case-02 | pass→pass | 23,672 | 32,995 | +39% | 1 | 1 | 0% | 2,330 | 3,556 | +53% | 0 | 0 | — |
case-03 | pass→fail | 25,455 | 32,009 | +26% | 1 | 1 | 0% | 2,225 | 5,170 | +132% | 0 | 0 | — |
case-04 | pass→pass | 24,578 | 8,288 | -66% | 1 | 1 | 0% | 898 | 1,275 | +42% | 0 | 0 | — |
case-05 | pass→pass | 41,789 | 20,775 | -50% | 1 | 1 | 0% | 1,447 | 2,203 | +52% | 0 | 0 | — |
case-06 | fail→pass | 9,638 | 26,091 | +171% | 1 | 1 | 0% | 1,598 | 2,110 | +32% | 0 | 0 | — |
case-07 | fail→pass | 13,401 | 13,763 | +3% | 1 | 1 | 0% | 1,160 | 1,764 | +52% | 0 | 0 | — |
case-08 | pass→pass | 24,890 | 31,816 | +28% | 1 | 1 | 0% | 1,509 | 2,923 | +94% | 0 | 0 | — |
case-09 | pass→pass | 36,976 | 18,372 | -50% | 1 | 1 | 0% | 1,498 | 2,665 | +78% | 0 | 0 | — |
case-10 | pass→pass | 12,113 | 32,412 | +168% | 1 | 1 | 0% | 626 | 2,680 | +328% | 0 | 0 | — |
case-11 | fail→pass | 37,075 | 29,636 | -20% | 1 | 1 | 0% | 3,007 | 2,483 | -17% | 0 | 0 | — |
case-12 | fail→pass | 10,642 | 30,400 | +186% | 1 | 1 | 0% | 1,420 | 2,193 | +54% | 0 | 0 | — |
case-13 | pass→pass | 11,252 | 30,611 | +172% | 1 | 1 | 0% | 729 | 2,077 | +185% | 0 | 0 | — |
case-14 | fail→pass | 33,046 | 11,683 | -65% | 1 | 1 | 0% | 1,022 | 2,023 | +98% | 0 | 0 | — |
case-15 | fail→pass | 28,813 | 12,224 | -58% | 1 | 1 | 0% | 1,253 | 1,956 | +56% | 0 | 0 | — |
case-16 | pass→pass | 19,761 | 12,154 | -38% | 1 | 1 | 0% | 1,024 | 2,156 | +111% | 0 | 0 | — |
case-17 | pass→pass | 4,139 | 11,623 | +181% | 1 | 1 | 0% | 587 | 1,690 | +188% | 0 | 0 | — |
case-18 | pass→pass | 7,435 | 10,786 | +45% | 1 | 1 | 0% | 821 | 1,662 | +102% | 0 | 0 | — |
case-19 | pass→pass | 12,572 | 15,441 | +23% | 1 | 1 | 0% | 1,022 | 1,844 | +80% | 0 | 0 | — |
case-20 | pass→fail | 10,740 | 35,460 | +230% | 1 | 1 | 0% | 1,009 | 7,090 | +603% | 0 | 0 | — |
case-21 | pass→fail | 11,219 | 36,780 | +228% | 1 | 1 | 0% | 1,873 | 7,163 | +282% | 0 | 0 | — |
case-22 | pass→fail | 14,789 | 19,102 | +29% | 1 | 1 | 0% | 1,759 | 3,872 | +120% | 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. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 4 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.