Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Iterative convergence to a single answer through Classic Delphi, Modified Delphi, or Nominal Group Technique rounds.
.claude/skills/yogsoth-ai-convergence-distillation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 60% | 0% |
Purpose: Drive multiple independent perspectives toward a single shared answer through iterative anonymous feedback rounds. Each round narrows the distribution of judgments until consensus threshold is met or stability is reached.
When to use:
| Parameter | Constraint | |-----------|-----------| | Rounds | 2–4 (stop at consensus or stability) | | Perspectives | ≥4 independent | | Consensus threshold | ≥70% agreement or IQR ≤ 1 |
| Key | Type | Description | |-----|------|-------------| | question | string | The focal question being converged | | perspectives | array | List of perspective descriptions | | rounds | array | History of all round results | | current_round | integer | Current round number | | consensus_score | float | Latest consensus measurement | | stability | boolean | Whether scores stopped changing |
yamlconsensus_reached: true/false final_answer: <aggregated result> consensus_score: <float> rounds_completed: <int> dissent_record: <items that did not converge>
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | iterative-convergence-round | Execute one full Delphi round — collect judgments, distribute anonymous feedback, measure consensus, decide whether to continue. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | consensus-synthesis | Synthesize all rounds into a final consensus report documenting agreements, dissent, and process. |
<!-- 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 | 10,715 | 18,479 | +72% | 1 | 1 | 0% | 1,909 | 3,992 | +109% | 0 | 0 | — |
case-02 | fail→pass | 10,133 | 16,149 | +59% | 1 | 1 | 0% | 1,902 | 3,414 | +79% | 0 | 0 | — |
case-03 | pass→fail | 4,551 | 9,496 | +109% | 1 | 1 | 0% | 745 | 2,240 | +201% | 0 | 0 | — |
case-04 | pass→fail | 25,120 | 28,663 | +14% | 1 | 1 | 0% | 3,591 | 6,033 | +68% | 0 | 0 | — |
case-05 | pass→pass | 12,208 | 13,412 | +10% | 1 | 1 | 0% | 2,437 | 3,202 | +31% | 0 | 0 | — |
case-06 | pass→pass | 9,203 | 2,982 | -68% | 1 | 1 | 0% | 1,335 | 1,032 | -23% | 0 | 0 | — |
case-07 | pass→pass | 11,376 | 5,309 | -53% | 1 | 1 | 0% | 1,684 | 1,403 | -17% | 0 | 0 | — |
case-08 | pass→pass | 8,362 | 4,383 | -48% | 1 | 1 | 0% | 1,223 | 1,196 | -2% | 0 | 0 | — |
case-09 | fail→fail | 9,301 | 7,148 | -23% | 1 | 1 | 0% | 1,289 | 1,107 | -14% | 0 | 0 | — |
case-10 | pass→pass | 10,535 | 4,970 | -53% | 1 | 1 | 0% | 1,727 | 1,175 | -32% | 0 | 0 | — |
case-11 | pass→pass | 7,045 | 1,730 | -75% | 1 | 1 | 0% | 978 | 757 | -23% | 0 | 0 | — |
case-12 | pass→pass | 9,247 | 2,408 | -74% | 1 | 1 | 0% | 1,283 | 898 | -30% | 0 | 0 | — |
case-13 | fail→pass | 9,766 | 1,676 | -83% | 1 | 1 | 0% | 1,505 | 781 | -48% | 0 | 0 | — |
case-14 | fail→pass | 5,505 | 2,248 | -59% | 1 | 1 | 0% | 793 | 830 | +5% | 0 | 0 | — |
case-15 | fail→pass | 3,906 | 1,903 | -51% | 1 | 1 | 0% | 530 | 847 | +60% | 0 | 0 | — |
case-16 | fail→pass | 10,163 | 2,730 | -73% | 1 | 1 | 0% | 1,484 | 957 | -36% | 0 | 0 | — |
case-17 | fail→pass | 8,399 | 2,696 | -68% | 1 | 1 | 0% | 1,208 | 942 | -22% | 0 | 0 | — |
case-18 | fail→pass | 9,496 | 1,812 | -81% | 1 | 1 | 0% | 1,368 | 820 | -40% | 0 | 0 | — |
case-19 | pass→pass | 11,550 | 4,398 | -62% | 1 | 1 | 0% | 1,696 | 1,308 | -23% | 0 | 0 | — |
case-20 | pass→pass | 7,974 | 1,438 | -82% | 1 | 1 | 0% | 1,180 | 778 | -34% | 0 | 0 | — |
case-21 | fail→pass | 5,498 | 1,592 | -71% | 1 | 1 | 0% | 712 | 803 | +13% | 0 | 0 | — |
case-22 | fail→pass | 9,711 | 2,065 | -79% | 1 | 1 | 0% | 1,446 | 853 | -41% | 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 21 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 +36 percentage points is the difference between those two pass rates over the 21 comparable cases. 2 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.