Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strategy for multi-judge ranking aggregation using Condorcet, Schulze, Borda, Kemeny-Young, and Copeland methods to produce consensus rankings from diverse perspectives.
.claude/skills/yogsoth-ai-collective-adjudication/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 209% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -46% | 0% |
Aggregate rankings from multiple independent judges into a single consensus ranking. Handles disagreement detection, voting paradoxes, and produces transparent aggregation with disagreement maps.
| Resource | Allocation | |----------|-----------| | Judges/Perspectives | ≥3 independent evaluators | | Comparisons per judge | Complete or near-complete per judge | | Aggregation methods | ≥2 methods for robustness check | | Disagreement threshold | Flag pairs where judges disagree >40% |
yamlcandidates: [] perspectives: [] # judge identities/prompts ballots: [] # [{judge, ranking: [...]}] aggregation_results: {} # method → consensus_ranking disagreement_map: {} # pair → {agreement_rate, split} cycles: [] # Condorcet cycles if any method: "" # schulze | borda | kemeny-young | copeland
yamlconsensus_ranking: - {rank: 1, candidate: "...", wins: 8, copeland_score: 0.95} - {rank: 2, candidate: "...", wins: 7, copeland_score: 0.88} method: schulze judges: 5 condorcet_winner: "candidate_a" # or null if cycle disagreement_hotspots: - {pair: ["c", "d"], agreement: 0.4, split: "3:2"} cross_validation: {borda_agreement: 0.92, copeland_agreement: 0.96}
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | consistency-audit-loop | Detect preference cycles, localize inconsistent judgments, request corrections, and recompute ratings until consistency threshold is met. | | multi-judge-aggregation | Collect independent rankings from multiple judges, aggregate using social choice methods, and identify disagreement hotspots. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | ranking-synthesis | Produce the final ranking artifact from converged ratings and consistency report. |
<!-- 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 | 27,188 | 25,619 | -6% | 1 | 1 | 0% | 6,242 | 6,585 | +5% | 0 | 0 | — |
case-02 | fail→pass | 25,590 | 25,357 | -1% | 1 | 1 | 0% | 5,668 | 6,595 | +16% | 0 | 0 | — |
case-03 | fail→pass | 11,571 | 26,836 | +132% | 1 | 1 | 0% | 2,140 | 6,622 | +209% | 0 | 0 | — |
case-04 | pass→pass | 2,957 | 5,877 | +99% | 1 | 1 | 0% | 642 | 1,903 | +196% | 0 | 0 | — |
case-05 | pass→pass | 3,984 | 4,780 | +20% | 1 | 1 | 0% | 879 | 1,844 | +110% | 0 | 0 | — |
case-06 | pass→pass | 3,931 | 4,124 | +5% | 1 | 1 | 0% | 885 | 1,574 | +78% | 0 | 0 | — |
case-07 | fail→pass | 23,068 | 1,944 | -92% | 1 | 1 | 0% | 1,062 | 997 | -6% | 0 | 0 | — |
case-08 | fail→pass | 12,794 | 2,720 | -79% | 1 | 1 | 0% | 2,050 | 1,105 | -46% | 0 | 0 | — |
case-09 | fail→pass | 13,948 | 2,772 | -80% | 1 | 1 | 0% | 2,353 | 1,254 | -47% | 0 | 0 | — |
case-10 | pass→pass | 11,978 | 3,478 | -71% | 1 | 1 | 0% | 1,844 | 1,258 | -32% | 0 | 0 | — |
case-11 | pass→pass | 13,140 | 7,563 | -42% | 1 | 1 | 0% | 2,711 | 2,182 | -20% | 0 | 0 | — |
case-12 | fail→pass | 12,301 | 3,082 | -75% | 1 | 1 | 0% | 2,156 | 1,168 | -46% | 0 | 0 | — |
case-13 | pass→pass | 15,476 | 12,440 | -20% | 1 | 1 | 0% | 2,379 | 2,950 | +24% | 0 | 0 | — |
case-14 | fail→pass | 13,098 | 9,157 | -30% | 1 | 1 | 0% | 2,198 | 2,218 | +1% | 0 | 0 | — |
case-15 | pass→pass | 7,316 | 5,068 | -31% | 1 | 1 | 0% | 1,138 | 1,501 | +32% | 0 | 0 | — |
case-16 | pass→pass | 13,694 | 14,063 | +3% | 1 | 1 | 0% | 2,682 | 3,549 | +32% | 0 | 0 | — |
case-17 | pass→pass | 17,544 | 13,925 | -21% | 1 | 1 | 0% | 2,840 | 2,910 | +2% | 0 | 0 | — |
case-18 | pass→pass | 5,481 | 2,995 | -45% | 1 | 1 | 0% | 839 | 1,324 | +58% | 0 | 0 | — |
case-19 | fail→pass | 10,372 | 8,875 | -14% | 1 | 1 | 0% | 1,762 | 2,200 | +25% | 0 | 0 | — |
case-20 | pass→pass | 17,518 | 11,467 | -35% | 1 | 1 | 0% | 3,093 | 2,813 | -9% | 0 | 0 | — |
case-21 | fail→pass | 5,145 | 3,153 | -39% | 1 | 1 | 0% | 760 | 1,252 | +65% | 0 | 0 | — |
case-22 | pass→pass | 16,015 | 12,887 | -20% | 1 | 1 | 0% | 2,588 | 3,106 | +20% | 0 | 0 | — |
case-23 | fail→fail | 14,349 | 1,787 | -88% | 1 | 1 | 0% | 2,176 | 980 | -55% | 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. 23 cases were attempted, and 22 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 +43 percentage points is the difference between those two pass rates over the 22 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.