Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Problem Reformulation Campaign — question the problem itself. Escape dominant ideas, reframe from multiple perspectives, apply dialectical inquiry, assess wickedness, discover appreciative alternatives. 5 strategies, 3 tactics, 10 subagent SOPs.
.claude/skills/yogsoth-ai-problem-reformulation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-03 | ✓→✗ | ▼ Worse | 184% | 0% |
| case-04 | ✓→✗ | ▼ Worse | -54% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -62% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 40% | 0% |
Question the problem itself — are we solving the right problem?
| Signal | Strategy | |--------|----------| | Dominant idea, paradigm lock-in, lateral thinking, escape | → dominant-idea-escape | | Multiple perspectives, CATWOE, reframing matrix, Rich Pictures | → multi-perspective-reframing | | Double-loop learning, governing variables, dialectical inquiry | → dialectical-reformulation | | Wicked problems, Rittel's criteria, complexity, Cynefin | → wickedness-assessment | | Positive deviance, Appreciative Inquiry, asset-based perspective | → appreciative-reframing |
lateral-escape, multi-worldview-comparison, dialectical-escalation
Import: web-search, web-research, paper-overview, paper-search, paper-research Subagent: dominant-idea-identification, provocation-generation, consequence-following, catwoe-analysis, reframing-matrix, governing-variable-surfacing, counter-assumption-generation, wickedness-scoring, appreciative-discovery, reformulation-synthesis Shared: assumption-surfacing, multi-stakeholder-simulation
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Strategy | When to use | | --- | --- | | appreciative-reframing | Find positive deviants and reframe the problem from deficit-based to asset-based using Appreciative Inquiry. | | dialectical-reformulation | Surface Argyris governing variables and test whether the problem dissolves under alternative governing variables (double-loop learning). | | dominant-idea-escape | Identify dominant paradigms constraining the field and use de Bono lateral thinking provocations to escape them. | | multi-perspective-reframing | Apply CATWOE from multiple stakeholder viewpoints and reframing matrix to reveal aspects invisible from the dominant perspective. | | wickedness-assessment | Apply Rittel's 10 criteria to determine if the problem is tame, complex, or wicked, and adjust research strategy accordingly. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | context-checkpoint | Append research process and results to the current Phase's context file. Covers both process and results with genuine substance. Use this skill at plan-designated checkpoint points — typically after each strategy completes or at key decision nodes within a research Phase. | | context-init | Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed. |
<!-- 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 | pass→pass | 24,314 | 29,868 | +23% | 1 | 1 | 0% | 3,157 | 4,433 | +40% | 0 | 0 | — |
case-02 | fail→pass | 41,772 | 57,387 | +37% | 1 | 1 | 0% | 5,795 | 8,298 | +43% | 0 | 0 | — |
case-03 | pass→fail | 21,825 | 48,768 | +123% | 1 | 1 | 0% | 2,823 | 8,014 | +184% | 0 | 0 | — |
case-09 | pass→pass | 17,405 | 30,837 | +77% | 1 | 1 | 0% | 2,639 | 6,038 | +129% | 0 | 0 | — |
case-04 | pass→fail | 25,535 | 25,161 | -1% | 1 | 1 | 0% | 3,454 | 1,598 | -54% | 0 | 0 | — |
case-05 | pass→pass | 26,040 | 42,775 | +64% | 1 | 1 | 0% | 2,973 | 7,209 | +142% | 0 | 0 | — |
case-06 | pass→pass | 19,247 | 18,741 | -3% | 1 | 1 | 0% | 3,018 | 3,622 | +20% | 0 | 0 | — |
case-07 | pass→pass | 22,513 | 32,085 | +43% | 1 | 1 | 0% | 2,720 | 4,803 | +77% | 0 | 0 | — |
case-08 | pass→pass | 23,056 | 36,408 | +58% | 1 | 1 | 0% | 2,807 | 4,051 | +44% | 0 | 0 | — |
case-10 | pass→pass | 24,587 | 27,059 | +10% | 1 | 1 | 0% | 2,782 | 4,350 | +56% | 0 | 0 | — |
case-11 | pass→pass | 25,064 | 31,715 | +27% | 1 | 1 | 0% | 3,069 | 4,596 | +50% | 0 | 0 | — |
case-12 | pass→pass | 20,191 | 44,169 | +119% | 1 | 1 | 0% | 3,185 | 6,572 | +106% | 0 | 0 | — |
case-13 | pass→pass | 31,798 | 29,280 | -8% | 1 | 1 | 0% | 3,265 | 4,716 | +44% | 0 | 0 | — |
case-14 | pass→pass | 29,534 | 42,473 | +44% | 1 | 1 | 0% | 3,656 | 4,394 | +20% | 0 | 0 | — |
case-15 | pass→pass | 22,905 | 37,128 | +62% | 1 | 1 | 0% | 2,474 | 5,954 | +141% | 0 | 0 | — |
case-16 | pass→pass | 17,519 | 48,886 | +179% | 1 | 1 | 0% | 1,639 | 3,416 | +108% | 0 | 0 | — |
case-17 | pass→pass | 13,277 | 36,064 | +172% | 1 | 1 | 0% | 1,942 | 6,428 | +231% | 0 | 0 | — |
case-18 | pass→fail | 21,081 | 28,886 | +37% | 1 | 1 | 0% | 3,233 | 1,221 | -62% | 0 | 0 | — |
case-19 | pass→pass | 66,259 | 53,769 | -19% | 1 | 1 | 0% | 3,474 | 7,893 | +127% | 0 | 0 | — |
case-20 | fail→fail | 22,325 | 33,437 | +50% | 1 | 1 | 0% | 1,730 | 912 | -47% | 0 | 0 | — |
case-21 | fail→fail | 16,297 | 37,921 | +133% | 1 | 1 | 0% | 2,686 | 3,683 | +37% | 0 | 0 | — |
case-22 | pass→pass | 20,605 | 22,226 | +8% | 1 | 1 | 0% | 2,988 | 3,822 | +28% | 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 20 counted toward the lift figure. The other 2 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 -9 percentage points is the difference between those two pass rates over the 20 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.