Install any skill in seconds. Free to start, no credit card required.
Get Started Free →What is the story of each future? — Shell method narrative construction for rich qualitative scenario understanding
.claude/skills/yogsoth-ai-narrative-scenario/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 105% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -3% | 0% |
Shell Scenario Method (Wack/van der Heijden). Construct rich, internally consistent narratives that illuminate qualitatively different futures. Focus on storytelling that reveals causal mechanisms and decision points rather than exhaustive enumeration.
Key principles:
scenario-driver-identificationscenario-narrative-construction (×4)scenario-impact-assessment (×4)robustness-scoringscenario-synthesis| Step | Token Budget | Notes | |------|-------------|-------| | Driver identification | 8K | PESTEL scan | | Axis selection | 4K | Ranking and selection | | Narrative construction | 15K × 4 | Rich storytelling | | Impact assessment | 10K × 4 | Per scenario | | Robustness scoring | 8K | Aggregation | | Synthesis | 12K | Final compilation |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | cross-consistency-filtering | Orchestrates pairwise consistency evaluation and narrative construction to filter the morphological field | | strategy-robustness-testing | Orchestrates impact assessment and robustness scoring to evaluate research approach resilience across scenarios |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | robustness-scoring | Compute robustness index across scenarios with sensitivity analysis | | scenario-driver-identification | Identify key uncertainty drivers using PESTEL framework scanning | | scenario-impact-assessment | Assess each scenario's impact on the research approach across multiple dimensions | | scenario-narrative-construction | Build rich narratives for surviving morphological configurations using Shell method | | scenario-synthesis | Comprehensive scenario analysis report synthesizing all scenario work |
<!-- 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 | 54,541 | 52,637 | -3% | 1 | 1 | 0% | 8,280 | 8,252 | -0% | 0 | 0 | — |
case-02 | fail→fail | 56,091 | 64,497 | +15% | 1 | 1 | 0% | 8,277 | 8,522 | +3% | 0 | 0 | — |
case-03 | fail→fail | 33,479 | 86,635 | +159% | 1 | 1 | 0% | 3,713 | 8,933 | +141% | 0 | 0 | — |
case-04 | fail→pass | 38,727 | 29,097 | -25% | 1 | 1 | 0% | 2,814 | 4,432 | +57% | 0 | 0 | — |
case-05 | pass→pass | 86,316 | 46,150 | -47% | 1 | 1 | 0% | 7,278 | 8,924 | +23% | 0 | 0 | — |
case-06 | pass→pass | 35,538 | 18,807 | -47% | 1 | 1 | 0% | 2,853 | 2,327 | -18% | 0 | 0 | — |
case-12 | fail→pass | 12,486 | 9,299 | -26% | 1 | 1 | 0% | 525 | 1,078 | +105% | 0 | 0 | — |
case-07 | fail→fail | 50,290 | 38,654 | -23% | 1 | 1 | 0% | 2,434 | 4,013 | +65% | 0 | 0 | — |
case-08 | fail→pass | 44,368 | 47,757 | +8% | 1 | 1 | 0% | 2,264 | 3,875 | +71% | 0 | 0 | — |
case-09 | fail→pass | 45,415 | 27,143 | -40% | 1 | 1 | 0% | 889 | 1,006 | +13% | 0 | 0 | — |
case-10 | fail→pass | 22,602 | 9,712 | -57% | 1 | 1 | 0% | 1,010 | 983 | -3% | 0 | 0 | — |
case-11 | fail→pass | 13,383 | 7,275 | -46% | 1 | 1 | 0% | 1,080 | 1,033 | -4% | 0 | 0 | — |
case-13 | fail→pass | 15,051 | 3,185 | -79% | 1 | 1 | 0% | 1,524 | 1,109 | -27% | 0 | 0 | — |
case-14 | fail→pass | 13,541 | 7,172 | -47% | 1 | 1 | 0% | 1,344 | 999 | -26% | 0 | 0 | — |
case-15 | fail→pass | 13,897 | 8,450 | -39% | 1 | 1 | 0% | 1,439 | 1,172 | -19% | 0 | 0 | — |
case-16 | fail→pass | 20,236 | 8,576 | -58% | 1 | 1 | 0% | 1,418 | 1,176 | -17% | 0 | 0 | — |
case-17 | pass→pass | 11,094 | 3,309 | -70% | 1 | 1 | 0% | 1,655 | 1,117 | -33% | 0 | 0 | — |
case-18 | fail→pass | 21,544 | 7,734 | -64% | 1 | 1 | 0% | 872 | 1,013 | +16% | 0 | 0 | — |
case-19 | fail→pass | 10,898 | 4,474 | -59% | 1 | 1 | 0% | 878 | 1,035 | +18% | 0 | 0 | — |
case-20 | pass→pass | 18,159 | 14,912 | -18% | 1 | 1 | 0% | 1,837 | 2,166 | +18% | 0 | 0 | — |
case-21 | pass→pass | 13,495 | 20,063 | +49% | 1 | 1 | 0% | 1,267 | 1,887 | +49% | 0 | 0 | — |
case-22 | fail→pass | 13,667 | 15,544 | +14% | 1 | 1 | 0% | 2,139 | 2,467 | +15% | 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 +59 percentage points is the difference between those two pass rates over the 20 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.