What problem does it solve? Maintaining a multi-volume knowledge base risks broken prerequisite references, dependency cycles, and silent contradictions between volumes that no single author notices. This Skill runs a two-stage audit that catches both structural errors and semantic inconsistencies before they propagate. ## Core Features & Use Cases - Deterministic gate 3 validation: Runs python -m ferramentas.validar --cross-refs to detect nonexistent depende_de ids and cycles in the prerequisite graph, with concrete fix guidance for each violation type. - Semantic contradiction audit: Dispatches the auditor-fable subagent to find duplicated term definitions, conflicting architectural decisions, divergent numeric values, and semantically inverted prerequisites across volumes. - Ordered two-phase workflow: Enforces that the deterministic check passes before the semantic pass runs, preventing analysis of a broken graph. - Use Case: After editing several _VOLUME.yml files in the AI-ENGINEERING-OS collection, run the cross-reference check to confirm no dependency cycles were introduced and that no two volumes now contradict each other. ## Quick Start Ask the assistant to run the cross-reference check on the AI-ENGINEERING-OS collection to validate the dependency graph and audit volumes for contradictions.
