What problem does it solve? Maintaining a large research knowledge base leads to undetected factual contradictions, stale versions, and silent query failures. This Skill performs content-level audits of AKNE knowledge base files (detecting contradictions, version clusters, research gaps, and assumptions) and provides structured KnowledgeGraph/QueryEngine access with explicit error handling instead of silent empty results. ## Core Features & Use Cases - Content-Level Auditing: Cross-compares physical parameter declarations across knowledge base source files and reports each contradiction with file path, declared value, correct-value source, and severity level. - KnowledgeGraph Queries: Resolves entities and traverses relations through QueryEngine APIs such as resolve_entity and get_relations, with fuzzy matching support. - Boundary Error Handling: Converts empty or unknown entity queries into structured errors with context and recovery hints rather than silent None results. - Use Case: Given BPPV knowledge base files, the audit detects that a declared otolith radius of 0.5-15 nm contradicts the anatomical standard of 5-30 μm for calcium carbonate crystals (a 1000x magnitude gap) and flags it as a fatal-severity contradiction with full traceability. ## Quick Start Audit the BPPV knowledge base source files for parameter contradictions and report each conflict with its file path, declared value, and correct-value source.
