Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply Hegelian dialectics (thesis-antithesis-synthesis) to analyze contradictions and generate higher-order understanding. Use this skill when the user needs to resolve opposing viewpoints, find creative synthesis between conflicting positions, or move past binary thinking — even if they say 'both sides have valid points', 'how do we reconcile these views', or 'is there a third option'.
.claude/skills/asgard-ai-platform-hum-dialectics/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-14 | ✓→✗ | ▼ Worse | 11% | 0% |
Dialectics is a method of reasoning through contradictions. Instead of choosing between opposing positions (thesis vs antithesis), it seeks a synthesis that transcends both — preserving what's valid in each while resolving the contradiction at a higher level.
IRON LAW: Synthesis ≠ Compromise
Synthesis is NOT "meeting in the middle." It's a NEW position that resolves
the contradiction by reframing the problem at a higher level of abstraction.
Compromise: "Let's do half of A and half of B"
Synthesis: "The contradiction between A and B reveals that C is the real answer"markdown# Dialectical Analysis: {Topic} ## Thesis {Position A — steelmanned} - What it gets right: ... - Where it fails: ... ## Antithesis {Position B — steelmanned} - What it gets right: ... - Where it fails: ... ## The Contradiction {What specific tension exists between these positions} ## Synthesis {Higher-order position that resolves the contradiction} - Preserves from thesis: ... - Preserves from antithesis: ... - Transcends by: ...
Scenario: "Should companies prioritize shareholder value or stakeholder value?"
references/marxist-dialectics.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 21,069 | 24,670 | +17% | 1 | 1 | 0% | 3,154 | 3,521 | +12% | 0 | 0 | — |
case-02 | fail→pass | 18,807 | 21,851 | +16% | 1 | 1 | 0% | 2,835 | 4,016 | +42% | 0 | 0 | — |
case-03 | fail→pass | 16,328 | 18,372 | +13% | 1 | 1 | 0% | 2,512 | 3,371 | +34% | 0 | 0 | — |
case-04 | fail→fail | 20,109 | 20,686 | +3% | 1 | 1 | 0% | 3,127 | 3,875 | +24% | 0 | 0 | — |
case-05 | pass→pass | 15,734 | 16,482 | +5% | 1 | 1 | 0% | 2,479 | 3,262 | +32% | 0 | 0 | — |
case-06 | pass→pass | 29,024 | 17,047 | -41% | 1 | 1 | 0% | 4,123 | 3,467 | -16% | 0 | 0 | — |
case-07 | pass→pass | 17,200 | 17,322 | +1% | 1 | 1 | 0% | 2,590 | 3,437 | +33% | 0 | 0 | — |
case-08 | pass→pass | 15,573 | 14,991 | -4% | 1 | 1 | 0% | 2,548 | 3,166 | +24% | 0 | 0 | — |
case-09 | fail→pass | 19,725 | 15,921 | -19% | 1 | 1 | 0% | 2,918 | 3,305 | +13% | 0 | 0 | — |
case-10 | fail→fail | 20,739 | 19,285 | -7% | 1 | 1 | 0% | 3,354 | 3,702 | +10% | 0 | 0 | — |
case-11 | fail→fail | 18,358 | 21,245 | +16% | 1 | 1 | 0% | 2,652 | 3,763 | +42% | 0 | 0 | — |
case-12 | pass→pass | 19,985 | 19,019 | -5% | 1 | 1 | 0% | 2,781 | 3,097 | +11% | 0 | 0 | — |
case-13 | pass→pass | 18,715 | 25,513 | +36% | 1 | 1 | 0% | 2,896 | 3,965 | +37% | 0 | 0 | — |
case-14 | pass→fail | 23,953 | 19,665 | -18% | 1 | 1 | 0% | 3,345 | 3,729 | +11% | 0 | 0 | — |
case-15 | pass→pass | 24,025 | 26,624 | +11% | 1 | 1 | 0% | 3,629 | 4,432 | +22% | 0 | 0 | — |
case-21 | pass→fail | 14,826 | 26,776 | +81% | 1 | 1 | 0% | 2,170 | 5,635 | +160% | 0 | 0 | — |
case-16 | fail→fail | 29,803 | 21,970 | -26% | 1 | 1 | 0% | 4,439 | 4,157 | -6% | 0 | 0 | — |
case-17 | pass→pass | 20,038 | 18,018 | -10% | 1 | 1 | 0% | 2,823 | 3,315 | +17% | 0 | 0 | — |
case-18 | pass→pass | 22,338 | 24,268 | +9% | 1 | 1 | 0% | 2,998 | 4,136 | +38% | 0 | 0 | — |
case-19 | pass→pass | 23,372 | 19,036 | -19% | 1 | 1 | 0% | 3,255 | 3,720 | +14% | 0 | 0 | — |
case-20 | pass→fail | 12,335 | 22,146 | +80% | 1 | 1 | 0% | 1,975 | 4,518 | +129% | 0 | 0 | — |
case-22 | fail→fail | 10,438 | 17,268 | +65% | 1 | 1 | 0% | 1,652 | 3,710 | +125% | 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 +5 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.