Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build logical causal trees from symptoms to root causes — list UDEs, connect causal chains, validate logic, locate root causes. Combines ishikawa-decomposition, current-reality-tree, and clr-validation SOPs.
.claude/skills/yogsoth-ai-causal-tree-building/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 47% | 0% |
Build formal causal trees from symptoms to root causes.
Subagent: five-whys-drilling, ishikawa-decomposition, current-reality-tree, clr-validation Import: paper-search
Decompose via Ishikawa (6M categories adapted for research: Methodology, Data, Theory, Measurement, Researchers, Environment). Build formal CRT with sufficient-cause logic. Validate every causal link with CLR 8-check.
<HARD-GATE>
- ishikawa diagrams: >= 1
- CRT constructed: >= 1
- CLR validations: >= 3 links validated
- root causes identified: >= 1
</HARD-GATE><!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | clr-validation | Apply Goldratt's 8 Categories of Legitimate Reservation to validate causal claims. Tests clarity, existence, sufficiency, and logical integrity. | | current-reality-tree | Build TOC Current Reality Trees — connect Undesirable Effects via sufficient-cause logic to identify 1-3 root causes. | | deep-insight-paper-search | AI-powered paper summary and search. Import of literature-engine/literature-search skill. AI summary level — cite as "AI-extracted" not "paper states". | | five-whys-drilling | Iterative "Why?" questioning (5+ levels) to drill from surface phenomenon to actionable root cause. Each level verified against evidence. | | ishikawa-decomposition | Decompose problems into 6M categories (Methodology, Data, Theory, Measurement, Researchers, Environment) via fishbone diagram analysis. |
<!-- 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-16 | fail→pass | 21,839 | 30,627 | +40% | 1 | 1 | 0% | 3,168 | 5,508 | +74% | 0 | 0 | — |
case-01 | fail→pass | 35,470 | 32,783 | -8% | 1 | 1 | 0% | 6,221 | 5,893 | -5% | 0 | 0 | — |
case-02 | fail→pass | 27,216 | 36,499 | +34% | 1 | 1 | 0% | 4,374 | 6,659 | +52% | 0 | 0 | — |
case-03 | fail→pass | 27,827 | 35,909 | +29% | 1 | 1 | 0% | 4,408 | 6,669 | +51% | 0 | 0 | — |
case-04 | pass→fail | 27,108 | 37,746 | +39% | 1 | 1 | 0% | 4,305 | 6,640 | +54% | 0 | 0 | — |
case-05 | pass→fail | 21,295 | 34,273 | +61% | 1 | 1 | 0% | 4,027 | 6,634 | +65% | 0 | 0 | — |
case-06 | pass→fail | 18,629 | 35,396 | +90% | 1 | 1 | 0% | 3,122 | 6,644 | +113% | 0 | 0 | — |
case-07 | fail→pass | 15,232 | 20,118 | +32% | 1 | 1 | 0% | 2,516 | 3,697 | +47% | 0 | 0 | — |
case-08 | fail→pass | 11,688 | 20,603 | +76% | 1 | 1 | 0% | 1,932 | 3,801 | +97% | 0 | 0 | — |
case-09 | pass→pass | 20,767 | 22,740 | +10% | 1 | 1 | 0% | 3,578 | 4,324 | +21% | 0 | 0 | — |
case-10 | pass→pass | 21,934 | 26,634 | +21% | 1 | 1 | 0% | 3,626 | 4,670 | +29% | 0 | 0 | — |
case-11 | fail→fail | 19,266 | 37,015 | +92% | 1 | 1 | 0% | 3,102 | 6,632 | +114% | 0 | 0 | — |
case-12 | fail→pass | 29,684 | 33,814 | +14% | 1 | 1 | 0% | 4,479 | 5,954 | +33% | 0 | 0 | — |
case-13 | fail→pass | 20,013 | 26,891 | +34% | 1 | 1 | 0% | 3,038 | 4,926 | +62% | 0 | 0 | — |
case-14 | pass→pass | 18,724 | 26,624 | +42% | 1 | 1 | 0% | 3,285 | 5,070 | +54% | 0 | 0 | — |
case-15 | pass→pass | 14,846 | 27,407 | +85% | 1 | 1 | 0% | 2,542 | 5,250 | +107% | 0 | 0 | — |
case-17 | fail→pass | 14,903 | 31,569 | +112% | 1 | 1 | 0% | 2,418 | 5,488 | +127% | 0 | 0 | — |
case-18 | fail→fail | 21,127 | 30,714 | +45% | 1 | 1 | 0% | 2,991 | 5,196 | +74% | 0 | 0 | — |
case-19 | pass→pass | 26,143 | 41,387 | +58% | 1 | 1 | 0% | 3,910 | 6,623 | +69% | 0 | 0 | — |
case-20 | pass→pass | 25,486 | 21,523 | -16% | 1 | 1 | 0% | 2,540 | 3,709 | +46% | 0 | 0 | — |
case-21 | fail→pass | 14,750 | 21,973 | +49% | 1 | 1 | 0% | 2,302 | 3,692 | +60% | 0 | 0 | — |
case-22 | pass→pass | 14,809 | 30,240 | +104% | 1 | 1 | 0% | 2,180 | 4,867 | +123% | 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 +32 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.