Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →First eliminate non-qualifying candidates with non-compensatory rules, then score survivors with full MCDA methods.
.claude/skills/yogsoth-ai-screening-then-scoring/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-25 | ✓→✗ | ▼ Worse | 27% | 0% |
First eliminate non-qualifying alternatives using non-compensatory rules, then perform fine-grained scoring and ranking of survivors. Suitable for scenarios with large candidate sets or hard constraints.
Elimination rationale + survivor ranking
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | alternative-scoring | Score each candidate alternative against all criteria to produce a score matrix. | | conjunctive-filter | Apply conjunctive screening rules to eliminate candidates that fail any threshold. | | criterion-definition | Extract evaluation criteria from research goals and candidate alternatives. | | dominance-check | Identify dominated and non-dominated alternatives in a score matrix using Pareto dominance. | | normalization | Normalize a score matrix using a specified method to make scores comparable across criteria. | | scoring-synthesis | Synthesize score matrix, rankings, and sensitivity analysis into a final recommendation. | | threshold-setting | Define minimum acceptable thresholds for each criterion based on context and constraints. | | weight-elicitation-sop | Compute criteria weights using a specified elicitation method (AHP, Swing, BWM, MACBETH, or Simos). |
<!-- 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-12 | pass→pass | 9,098 | 13,197 | +45% | 1 | 1 | 0% | 1,504 | 1,926 | +28% | 0 | 0 | — |
case-01 | pass→pass | 14,555 | 16,261 | +12% | 1 | 1 | 0% | 2,288 | 2,397 | +5% | 0 | 0 | — |
case-02 | fail→pass | 9,979 | 8,311 | -17% | 1 | 1 | 0% | 1,592 | 1,843 | +16% | 0 | 0 | — |
case-03 | pass→pass | 17,952 | 15,522 | -14% | 1 | 1 | 0% | 1,889 | 2,192 | +16% | 0 | 0 | — |
case-04 | fail→pass | 11,688 | 11,096 | -5% | 1 | 1 | 0% | 1,172 | 1,526 | +30% | 0 | 0 | — |
case-05 | fail→fail | 10,746 | 8,408 | -22% | 1 | 1 | 0% | 822 | 1,138 | +38% | 0 | 0 | — |
case-06 | pass→pass | 13,801 | 11,706 | -15% | 1 | 1 | 0% | 1,374 | 1,699 | +24% | 0 | 0 | — |
case-07 | fail→pass | 20,266 | 8,730 | -57% | 1 | 1 | 0% | 2,354 | 2,043 | -13% | 0 | 0 | — |
case-08 | fail→fail | 10,051 | 8,792 | -13% | 1 | 1 | 0% | 812 | 1,128 | +39% | 0 | 0 | — |
case-09 | pass→pass | 9,950 | 9,581 | -4% | 1 | 1 | 0% | 801 | 1,341 | +67% | 0 | 0 | — |
case-10 | pass→pass | 16,344 | 14,942 | -9% | 1 | 1 | 0% | 1,643 | 2,169 | +32% | 0 | 0 | — |
case-11 | pass→pass | 28,860 | 8,347 | -71% | 1 | 1 | 0% | 1,402 | 1,075 | -23% | 0 | 0 | — |
case-13 | fail→fail | 8,097 | 11,858 | +46% | 1 | 1 | 0% | 1,266 | 1,676 | +32% | 0 | 0 | — |
case-14 | fail→fail | 10,009 | 8,061 | -19% | 1 | 1 | 0% | 1,552 | 2,059 | +33% | 0 | 0 | — |
case-15 | pass→pass | 17,777 | 14,559 | -18% | 1 | 1 | 0% | 2,163 | 2,198 | +2% | 0 | 0 | — |
case-16 | pass→pass | 16,072 | 11,323 | -30% | 1 | 1 | 0% | 1,587 | 1,648 | +4% | 0 | 0 | — |
case-17 | fail→fail | 9,464 | 3,717 | -61% | 1 | 1 | 0% | 687 | 1,138 | +66% | 0 | 0 | — |
case-18 | pass→pass | 10,969 | 7,809 | -29% | 1 | 1 | 0% | 1,645 | 1,861 | +13% | 0 | 0 | — |
case-19 | pass→pass | 16,040 | 13,627 | -15% | 1 | 1 | 0% | 1,738 | 1,880 | +8% | 0 | 0 | — |
case-20 | fail→pass | 8,512 | 11,772 | +38% | 1 | 1 | 0% | 1,463 | 1,766 | +21% | 0 | 0 | — |
case-21 | fail→fail | 12,277 | 10,939 | -11% | 1 | 1 | 0% | 1,160 | 1,587 | +37% | 0 | 0 | — |
case-22 | pass→pass | 5,798 | 5,818 | +0% | 1 | 1 | 0% | 935 | 1,515 | +62% | 0 | 0 | — |
case-23 | pass→pass | 20,515 | 17,853 | -13% | 1 | 1 | 0% | 2,191 | 2,663 | +22% | 0 | 0 | — |
case-24 | pass→pass | 23,900 | 16,688 | -30% | 1 | 1 | 0% | 1,925 | 3,409 | +77% | 0 | 0 | — |
case-25 | pass→fail | 26,023 | 27,134 | +4% | 1 | 1 | 0% | 3,388 | 4,316 | +27% | 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. 25 cases were attempted. The headline lift of +12 percentage points is the difference between those two pass rates over the 25 comparable cases. 1 case got worse with the skill loaded, and it is 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.