Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Double-loop learning escalation — surface governing variables, generate counter-assumptions, test if problem dissolves under alternatives, score wickedness if it persists.
.claude/skills/yogsoth-ai-deep-insight-dialectical-escalation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 205% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -3% | 0% |
Escalate from single-loop to double-loop learning.
governing-variable-surfacing → counter-assumption-generation → wickedness-scoring
Subagent: governing-variable-surfacing, counter-assumption-generation, wickedness-scoring Shared: assumption-surfacing
Surface the governing variables (Argyris: the unstated rules everyone follows), generate the opposite assumption for each, test whether the problem still exists under the alternative. If it persists regardless, assess wickedness level.
Single-loop: "How do we solve this problem better?" Double-loop: "Should we be solving this problem at all?"
<HARD-GATE>
- Governing variables surfaced: >= 3
- Counter-assumptions generated: >= 3
- Problem dissolution tests: >= 2
- Wickedness assessment: completed if problem persists
</HARD-GATE><!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | counter-assumption-generation | Generate dialectical opposites for governing variables — coherent alternative worldviews where the opposite is true. | | deep-insight-assumption-surfacing | Systematically extract implicit assumptions from methods, frameworks, or arguments. Identifies what is taken for granted without explicit justification. | | governing-variable-surfacing | Apply Argyris framework to identify governing variables — the unstated rules driving behavior in a research field. | | wickedness-scoring | Score a problem against Rittel's 10 criteria to determine if it is tame, complex, or wicked. |
<!-- 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-04 | pass→pass | 18,233 | 23,032 | +26% | 1 | 1 | 0% | 2,688 | 3,858 | +44% | 0 | 0 | — |
case-05 | pass→pass | 18,136 | 23,788 | +31% | 1 | 1 | 0% | 2,742 | 4,174 | +52% | 0 | 0 | — |
case-01 | pass→pass | 17,327 | 19,265 | +11% | 1 | 1 | 0% | 2,525 | 3,331 | +32% | 0 | 0 | — |
case-02 | fail→pass | 23,834 | 30,517 | +28% | 1 | 1 | 0% | 2,882 | 5,186 | +80% | 0 | 0 | — |
case-03 | pass→pass | 16,157 | 32,426 | +101% | 1 | 1 | 0% | 2,358 | 4,694 | +99% | 0 | 0 | — |
case-06 | pass→pass | 16,702 | 20,678 | +24% | 1 | 1 | 0% | 2,399 | 3,936 | +64% | 0 | 0 | — |
case-07 | fail→pass | 21,228 | 17,493 | -18% | 1 | 1 | 0% | 3,038 | 3,009 | -1% | 0 | 0 | — |
case-08 | pass→pass | 17,772 | 17,499 | -2% | 1 | 1 | 0% | 2,483 | 3,079 | +24% | 0 | 0 | — |
case-09 | pass→pass | 20,742 | 17,212 | -17% | 1 | 1 | 0% | 3,120 | 2,905 | -7% | 0 | 0 | — |
case-10 | fail→pass | 28,891 | 26,110 | -10% | 1 | 1 | 0% | 4,425 | 4,310 | -3% | 0 | 0 | — |
case-11 | pass→pass | 16,951 | 29,318 | +73% | 1 | 1 | 0% | 2,504 | 4,370 | +75% | 0 | 0 | — |
case-12 | pass→pass | 20,683 | 29,588 | +43% | 1 | 1 | 0% | 3,107 | 4,951 | +59% | 0 | 0 | — |
case-13 | fail→pass | 13,751 | 35,157 | +156% | 1 | 1 | 0% | 1,988 | 6,062 | +205% | 0 | 0 | — |
case-14 | pass→pass | 17,844 | 24,810 | +39% | 1 | 1 | 0% | 2,737 | 4,220 | +54% | 0 | 0 | — |
case-15 | fail→pass | 14,809 | 9,737 | -34% | 1 | 1 | 0% | 2,062 | 2,004 | -3% | 0 | 0 | — |
case-16 | fail→pass | 19,484 | 24,641 | +26% | 1 | 1 | 0% | 2,688 | 4,208 | +57% | 0 | 0 | — |
case-17 | pass→pass | 19,935 | 24,120 | +21% | 1 | 1 | 0% | 2,862 | 3,977 | +39% | 0 | 0 | — |
case-18 | pass→pass | 19,286 | 21,374 | +11% | 1 | 1 | 0% | 3,037 | 3,799 | +25% | 0 | 0 | — |
case-19 | fail→pass | 16,865 | 26,461 | +57% | 1 | 1 | 0% | 2,533 | 4,340 | +71% | 0 | 0 | — |
case-20 | pass→pass | 13,123 | 26,890 | +105% | 1 | 1 | 0% | 2,046 | 4,668 | +128% | 0 | 0 | — |
case-21 | pass→fail | 12,452 | 25,558 | +105% | 1 | 1 | 0% | 1,975 | 4,290 | +117% | 0 | 0 | — |
case-22 | pass→pass | 19,197 | 34,429 | +79% | 1 | 1 | 0% | 3,396 | 5,235 | +54% | 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 +27 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.