Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Map the structure of disagreement across perspectives using Policy Delphi, Argument Delphi, or SAST methods.
.claude/skills/yogsoth-ai-disagreement-cartography/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -9% | 0% |
Purpose: Rather than forcing convergence, map the topology of disagreement — identify clusters of opinion, the arguments supporting each cluster, and the fault lines between them. Useful for wicked problems where premature consensus would be harmful.
When to use:
| Parameter | Constraint | |-----------|-----------| | Rounds | 1–2 (mapping, not converging) | | Perspectives | ≥4 independent | | Cluster minimum | ≥2 distinct clusters to be useful |
| Key | Type | Description | |-----|------|-------------| | question | string | The focal question or issue | | perspectives | array | List of perspective descriptions | | judgments | array | Raw collected judgments | | clusters | array | Identified opinion clusters | | arguments | array | Extracted arguments per cluster | | disagreement_map | object | Final visualization structure |
yamlclusters: - id: <cluster_id> position: <summary> size: <number of perspectives> core_arguments: [...] fault_lines: - between: [cluster_a, cluster_b] nature: <empirical/value/definitional> bridging_possibilities: [...]
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | disagreement-mapping | Map disagreement structure by collecting judgments, clustering opinions, extracting arguments per cluster, and visualizing fault lines. |
<!-- 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→fail | 19,895 | 28,024 | +41% | 1 | 1 | 0% | 3,044 | 4,989 | +64% | 0 | 0 | — |
case-02 | fail→fail | 23,372 | 22,487 | -4% | 1 | 1 | 0% | 3,415 | 3,990 | +17% | 0 | 0 | — |
case-03 | fail→fail | 20,889 | 16,786 | -20% | 1 | 1 | 0% | 3,115 | 2,995 | -4% | 0 | 0 | — |
case-04 | pass→pass | 19,779 | 21,590 | +9% | 1 | 1 | 0% | 3,033 | 3,829 | +26% | 0 | 0 | — |
case-05 | fail→pass | 24,847 | 17,064 | -31% | 1 | 1 | 0% | 3,717 | 3,178 | -15% | 0 | 0 | — |
case-06 | fail→pass | 17,655 | 11,581 | -34% | 1 | 1 | 0% | 2,879 | 2,173 | -25% | 0 | 0 | — |
case-07 | pass→pass | 16,626 | 16,364 | -2% | 1 | 1 | 0% | 2,738 | 2,946 | +8% | 0 | 0 | — |
case-08 | fail→pass | 14,703 | 18,668 | +27% | 1 | 1 | 0% | 2,459 | 3,499 | +42% | 0 | 0 | — |
case-09 | fail→pass | 15,181 | 17,378 | +14% | 1 | 1 | 0% | 2,297 | 3,504 | +53% | 0 | 0 | — |
case-10 | fail→pass | 18,422 | 14,023 | -24% | 1 | 1 | 0% | 3,001 | 2,723 | -9% | 0 | 0 | — |
case-11 | fail→pass | 17,466 | 19,100 | +9% | 1 | 1 | 0% | 2,773 | 3,391 | +22% | 0 | 0 | — |
case-12 | fail→pass | 14,505 | 12,637 | -13% | 1 | 1 | 0% | 2,193 | 2,417 | +10% | 0 | 0 | — |
case-13 | fail→pass | 19,231 | 14,009 | -27% | 1 | 1 | 0% | 2,864 | 2,653 | -7% | 0 | 0 | — |
case-14 | fail→pass | 15,500 | 12,677 | -18% | 1 | 1 | 0% | 2,355 | 2,363 | +0% | 0 | 0 | — |
case-15 | fail→pass | 20,925 | 16,524 | -21% | 1 | 1 | 0% | 3,316 | 2,908 | -12% | 0 | 0 | — |
case-16 | fail→pass | 18,007 | 16,131 | -10% | 1 | 1 | 0% | 2,805 | 2,965 | +6% | 0 | 0 | — |
case-17 | fail→pass | 21,497 | 15,621 | -27% | 1 | 1 | 0% | 3,380 | 2,981 | -12% | 0 | 0 | — |
case-18 | fail→pass | 16,203 | 11,960 | -26% | 1 | 1 | 0% | 2,742 | 2,456 | -10% | 0 | 0 | — |
case-19 | pass→fail | 12,245 | 21,254 | +74% | 1 | 1 | 0% | 1,795 | 3,881 | +116% | 0 | 0 | — |
case-20 | pass→pass | 10,787 | 16,899 | +57% | 1 | 1 | 0% | 1,742 | 3,281 | +88% | 0 | 0 | — |
case-21 | pass→pass | 17,848 | 19,902 | +12% | 1 | 1 | 0% | 2,498 | 3,396 | +36% | 0 | 0 | — |
case-22 | pass→pass | 6,840 | 13,469 | +97% | 1 | 1 | 0% | 1,244 | 2,875 | +131% | 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 +55 percentage points is the difference between those two pass rates over the 22 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.