Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Construct the strongest possible counter-argument to the convergence decision using Dialectical Inquiry and Thesis-Antithesis-Synthesis methods.
.claude/skills/yogsoth-ai-counter-thesis-construction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 89% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 37% | 0% |
Purpose: Build the most compelling counter-thesis to the convergence winner — not to overturn the decision, but to ensure it can withstand the strongest possible intellectual challenge and to surface any synthesis opportunities.
When to use:
| Metric | Minimum | |--------|---------| | Counter-thesis depth | Full argument with evidence | | Debate rounds | >= 2 (thesis vs antithesis) | | Synthesis attempts | >= 1 |
yamlthesis: <the convergence decision> antithesis: <constructed counter-argument> debate_rounds: [] synthesis_attempted: false synthesis_result: null final_verdict: null # THESIS_STANDS | ANTITHESIS_WINS | SYNTHESIS_FOUND
| Tactic | When to Deploy | |--------|---------------| | adversarial-debate-protocol | Default — formal advocate/critic/judge cycle | | assumption-excavation | When counter-thesis rests on different assumptions |
yamlstrategy: counter-thesis-construction thesis: <decision> antithesis: <counter-argument> antithesis_strength: <1-10> debate_rounds: <count> synthesis_found: true | false synthesis: <if found> verdict: THESIS_STANDS | ANTITHESIS_WINS | SYNTHESIS_FOUND reasoning: <key evidence> conditions: []
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | adversarial-debate-protocol | Structured debate protocol that constructs an advocate, deploys critic attacks, and renders a judge verdict through iterative rounds. | | assumption-excavation | Systematic extraction, challenge, and sensitivity analysis of assumptions underlying a decision to identify load-bearing beliefs. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | steel-manning-synthesis | Synthesize all attacks and verdicts into a final unified assessment with surviving concerns and recommended modifications. |
<!-- 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-08 | fail→pass | 12,060 | 7,177 | -40% | 1 | 1 | 0% | 1,923 | 1,775 | -8% | 0 | 0 | — |
case-09 | fail→fail | 6,584 | 13,523 | +105% | 1 | 1 | 0% | 1,034 | 2,795 | +170% | 0 | 0 | — |
case-01 | fail→pass | 18,666 | 24,056 | +29% | 1 | 1 | 0% | 3,029 | 4,422 | +46% | 0 | 0 | — |
case-02 | fail→pass | 11,736 | 20,401 | +74% | 1 | 1 | 0% | 1,966 | 3,713 | +89% | 0 | 0 | — |
case-03 | fail→pass | 25,492 | 21,494 | -16% | 1 | 1 | 0% | 3,921 | 4,072 | +4% | 0 | 0 | — |
case-04 | pass→fail | 15,671 | 29,687 | +89% | 1 | 1 | 0% | 2,533 | 4,431 | +75% | 0 | 0 | — |
case-05 | pass→pass | 19,191 | 29,866 | +56% | 1 | 1 | 0% | 2,916 | 5,234 | +79% | 0 | 0 | — |
case-06 | pass→fail | 23,392 | 35,369 | +51% | 1 | 1 | 0% | 5,228 | 6,823 | +31% | 0 | 0 | — |
case-07 | fail→pass | 17,053 | 16,667 | -2% | 1 | 1 | 0% | 2,385 | 3,260 | +37% | 0 | 0 | — |
case-10 | fail→pass | 3,266 | 4,067 | +25% | 1 | 1 | 0% | 717 | 1,537 | +114% | 0 | 0 | — |
case-11 | fail→pass | 14,857 | 13,875 | -7% | 1 | 1 | 0% | 2,729 | 3,096 | +13% | 0 | 0 | — |
case-12 | fail→pass | 13,894 | 23,010 | +66% | 1 | 1 | 0% | 2,521 | 3,891 | +54% | 0 | 0 | — |
case-13 | fail→pass | 15,222 | 17,592 | +16% | 1 | 1 | 0% | 2,431 | 3,467 | +43% | 0 | 0 | — |
case-14 | fail→pass | 18,192 | 15,190 | -17% | 1 | 1 | 0% | 2,934 | 2,996 | +2% | 0 | 0 | — |
case-15 | fail→pass | 20,511 | 15,345 | -25% | 1 | 1 | 0% | 3,483 | 3,204 | -8% | 0 | 0 | — |
case-16 | fail→pass | 15,752 | 13,842 | -12% | 1 | 1 | 0% | 2,449 | 2,751 | +12% | 0 | 0 | — |
case-17 | fail→pass | 13,810 | 13,360 | -3% | 1 | 1 | 0% | 2,143 | 2,744 | +28% | 0 | 0 | — |
case-18 | fail→pass | 20,496 | 16,482 | -20% | 1 | 1 | 0% | 3,108 | 3,129 | +1% | 0 | 0 | — |
case-19 | fail→pass | 13,308 | 20,623 | +55% | 1 | 1 | 0% | 1,979 | 4,007 | +102% | 0 | 0 | — |
case-20 | fail→pass | 19,569 | 12,741 | -35% | 1 | 1 | 0% | 3,054 | 2,731 | -11% | 0 | 0 | — |
case-21 | fail→pass | 14,124 | 13,007 | -8% | 1 | 1 | 0% | 2,220 | 2,593 | +17% | 0 | 0 | — |
case-22 | fail→pass | 22,361 | 15,487 | -31% | 1 | 1 | 0% | 3,699 | 3,169 | -14% | 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 +73 percentage points is the difference between those two pass rates over the 22 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.