Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Campaign for building domain ontologies — systematic concept extraction, relation typing, taxonomy construction, and iterative refinement. Produces a structured concept hierarchy in the wiki vault.
.claude/skills/yogsoth-ai-ontology-building/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 111% | 0% |
Build a structured ontology for a research domain. Extracts concepts from sources, types their relationships, constructs taxonomies, validates consistency, and iteratively refines until the ontology is coherent and complete.
| Level | Count | Skills | |-------|-------|--------| | Strategy | 5 | domain-scoping, concept-extraction, relation-typing, taxonomy-validation, ontology-refinement | | Tactic | 3 | concept-decomposition, hierarchy-construction, consistency-checking | | SOP | 10 | seed-concept-search, source-gathering, concept-page-creation, alias-resolution, edge-batch-creation, hierarchy-visualization, gap-detection, merge-candidates, confidence-update, ontology-export |
| Metric | Small | Medium | Large | |--------|-------|--------|-------| | Source pages ingested | 10 | 25 | 50 | | Concept pages created | 15 | 40 | 80 | | Edges created | 30 | 100 | 200 | | Taxonomy depth (levels) | 2 | 3 | 4 | | Validation passes | 1 | 2 | 3 |
CC may reorder, skip, or repeat strategies based on the domain's needs.
vault_search — find existing concepts, detect duplicatesvault_add_edge — create typed relationshipsvault_query_graph — explore concept neighborhoodsvault_graph_stats — assess ontology coverage and connectivityvault_lint — validate structural integrityvault_index — maintain search index<HARD-GATE>
context-init at campaign startcontext-checkpoint after each strategy completesknowledge-compilation (wiki-vault tactic) after each strategy</HARD-GATE>
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Strategy | When to use | | --- | --- | | concept-extraction | Strategy for mining concepts from sources — systematic extraction, page creation, alias resolution. | | domain-scoping | Strategy for defining ontology boundaries — identify seed concepts, classify topic size, establish scope constraints. | | ontology-refinement | Strategy for iterative ontology improvement — merge duplicates, fill gaps, update confidence, prune dead branches. | | relation-typing | Strategy for identifying and typing relationships between concepts — create edges with appropriate types and weights. | | taxonomy-validation | Strategy for validating ontology consistency — check hierarchy, detect cycles, verify completeness. |
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | concept-decomposition | Tactic for breaking compound concepts into atomic parts — split over-broad concepts, identify sub-components, create child pages. | | hierarchy-construction | Tactic for building is-a and part-of hierarchies — establish parent-child relationships, verify transitivity, detect cycles. | | knowledge-compilation | Tactic for compiling research findings into vault pages — orchestrates page creation, updates, edge linking, and index maintenance. Minimum yield ≥3 page operations per invocation. | | 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 | | --- | --- | | 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 | fail→fail | 42,481 | 18,946 | -55% | 1 | 1 | 0% | 7,936 | 1,537 | -81% | 0 | 0 | — |
case-02 | fail→pass | 41,151 | 60,734 | +48% | 1 | 1 | 0% | 8,108 | 9,380 | +16% | 0 | 0 | — |
case-03 | fail→fail | 43,258 | 52,212 | +21% | 1 | 1 | 0% | 8,164 | 9,370 | +15% | 0 | 0 | — |
case-04 | pass→fail | 26,098 | 6,756 | -74% | 1 | 1 | 0% | 4,253 | 1,541 | -64% | 0 | 0 | — |
case-05 | pass→pass | 24,867 | 42,935 | +73% | 1 | 1 | 0% | 4,049 | 9,346 | +131% | 0 | 0 | — |
case-06 | fail→fail | 15,327 | 20,109 | +31% | 1 | 1 | 0% | 2,583 | 1,740 | -33% | 0 | 0 | — |
case-07 | pass→pass | 19,514 | 14,839 | -24% | 1 | 1 | 0% | 2,276 | 2,536 | +11% | 0 | 0 | — |
case-08 | pass→pass | 18,358 | 20,207 | +10% | 1 | 1 | 0% | 1,935 | 3,490 | +80% | 0 | 0 | — |
case-09 | pass→pass | 16,709 | 6,927 | -59% | 1 | 1 | 0% | 1,696 | 2,206 | +30% | 0 | 0 | — |
case-10 | fail→pass | 22,636 | 4,089 | -82% | 1 | 1 | 0% | 2,902 | 1,775 | -39% | 0 | 0 | — |
case-11 | fail→pass | 19,044 | 3,106 | -84% | 1 | 1 | 0% | 2,246 | 1,640 | -27% | 0 | 0 | — |
case-12 | fail→pass | 14,837 | 2,847 | -81% | 1 | 1 | 0% | 1,545 | 1,557 | +1% | 0 | 0 | — |
case-13 | fail→pass | 6,087 | 2,365 | -61% | 1 | 1 | 0% | 708 | 1,496 | +111% | 0 | 0 | — |
case-14 | pass→pass | 16,190 | 7,900 | -51% | 1 | 1 | 0% | 1,787 | 2,432 | +36% | 0 | 0 | — |
case-15 | fail→pass | 11,775 | 8,326 | -29% | 1 | 1 | 0% | 1,666 | 2,394 | +44% | 0 | 0 | — |
case-16 | fail→pass | 15,153 | 11,378 | -25% | 1 | 1 | 0% | 1,493 | 2,187 | +46% | 0 | 0 | — |
case-17 | fail→pass | 15,086 | 7,113 | -53% | 1 | 1 | 0% | 1,453 | 1,383 | -5% | 0 | 0 | — |
case-18 | pass→pass | 13,990 | 9,153 | -35% | 1 | 1 | 0% | 1,372 | 1,802 | +31% | 0 | 0 | — |
case-19 | pass→pass | 20,755 | 3,544 | -83% | 1 | 1 | 0% | 2,417 | 1,713 | -29% | 0 | 0 | — |
case-20 | pass→pass | 17,728 | 8,214 | -54% | 1 | 1 | 0% | 1,834 | 1,570 | -14% | 0 | 0 | — |
case-21 | pass→pass | 17,638 | 9,792 | -44% | 1 | 1 | 0% | 1,901 | 1,797 | -5% | 0 | 0 | — |
case-22 | pass→pass | 23,092 | 9,859 | -57% | 1 | 1 | 0% | 2,789 | 1,952 | -30% | 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 19 counted toward the lift figure. The other 3 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 +32 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 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.