Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Tactic for building is-a and part-of hierarchies — establish parent-child relationships, verify transitivity, detect cycles.
.claude/skills/yogsoth-ai-hierarchy-construction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -26% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 24% | 0% |
Build taxonomic hierarchies using component_of and instance_of edges. Ensure hierarchies are acyclic, transitive, and complete.
<HARD-GATE> ≥3 hierarchy edges created per invocation. If the hierarchy is already complete for the current scope, report and exit. </HARD-GATE>
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | knowledge-structuring-consistency-checking | Tactic for verifying ontology consistency — detect contradictions, cycles, orphans, and type violations. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | edge-batch-creation | SOP for creating multiple edges in a batch — efficient bulk relationship creation. | | gap-detection | SOP for finding structural gaps in the ontology — missing concepts, thin branches, disconnected clusters. | | hierarchy-visualization | SOP for inspecting the current hierarchy structure — query graph to display parent-child relationships. |
<!-- 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-02 | fail→fail | 19,678 | 3,174 | -84% | 1 | 1 | 0% | 4,031 | 647 | -84% | 0 | 0 | — |
case-01 | fail→pass | 23,395 | 22,332 | -5% | 1 | 1 | 0% | 4,133 | 3,056 | -26% | 0 | 0 | — |
case-03 | pass→pass | 20,116 | 23,349 | +16% | 1 | 1 | 0% | 3,395 | 4,589 | +35% | 0 | 0 | — |
case-04 | fail→fail | 14,121 | 15,701 | +11% | 1 | 1 | 0% | 2,248 | 3,006 | +34% | 0 | 0 | — |
case-05 | pass→pass | 11,074 | 8,107 | -27% | 1 | 1 | 0% | 1,910 | 1,859 | -3% | 0 | 0 | — |
case-06 | pass→pass | 8,121 | 9,514 | +17% | 1 | 1 | 0% | 1,221 | 1,982 | +62% | 0 | 0 | — |
case-07 | pass→pass | 12,803 | 12,566 | -2% | 1 | 1 | 0% | 1,924 | 2,460 | +28% | 0 | 0 | — |
case-08 | pass→pass | 9,752 | 13,034 | +34% | 1 | 1 | 0% | 1,365 | 2,508 | +84% | 0 | 0 | — |
case-09 | pass→fail | 6,361 | 11,888 | +87% | 1 | 1 | 0% | 1,180 | 2,423 | +105% | 0 | 0 | — |
case-10 | fail→pass | 9,239 | 13,983 | +51% | 1 | 1 | 0% | 1,549 | 2,733 | +76% | 0 | 0 | — |
case-11 | fail→pass | 11,752 | 14,190 | +21% | 1 | 1 | 0% | 1,941 | 2,810 | +45% | 0 | 0 | — |
case-12 | fail→pass | 8,534 | 7,857 | -8% | 1 | 1 | 0% | 1,541 | 1,809 | +17% | 0 | 0 | — |
case-13 | fail→pass | 10,651 | 11,265 | +6% | 1 | 1 | 0% | 1,932 | 2,404 | +24% | 0 | 0 | — |
case-14 | pass→pass | 7,488 | 10,004 | +34% | 1 | 1 | 0% | 1,133 | 1,867 | +65% | 0 | 0 | — |
case-15 | pass→pass | 8,886 | 11,463 | +29% | 1 | 1 | 0% | 1,376 | 2,288 | +66% | 0 | 0 | — |
case-16 | pass→pass | 6,421 | 11,412 | +78% | 1 | 1 | 0% | 1,044 | 2,159 | +107% | 0 | 0 | — |
case-17 | fail→pass | 11,688 | 13,119 | +12% | 1 | 1 | 0% | 1,875 | 2,574 | +37% | 0 | 0 | — |
case-18 | pass→pass | 10,630 | 7,733 | -27% | 1 | 1 | 0% | 1,848 | 1,824 | -1% | 0 | 0 | — |
case-19 | pass→pass | 3,359 | 5,410 | +61% | 1 | 1 | 0% | 624 | 1,343 | +115% | 0 | 0 | — |
case-20 | pass→pass | 9,486 | 8,262 | -13% | 1 | 1 | 0% | 1,746 | 1,815 | +4% | 0 | 0 | — |
case-21 | pass→pass | 13,064 | 15,302 | +17% | 1 | 1 | 0% | 2,025 | 2,265 | +12% | 0 | 0 | — |
case-22 | pass→fail | 8,141 | 6,016 | -26% | 1 | 1 | 0% | 1,298 | 896 | -31% | 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 +18 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.