What problem does it solve? AI models often hallucinate Wikidata entity IDs (Q-numbers) and property IDs (P-numbers) from training knowledge, producing wrong or empty query results. This Skill enforces a mandatory workflow that resolves every entity and property through live search before writing SPARQL, ensuring answers come from real Wikidata data rather than unreliable memorized knowledge. ## Core Features & Use Cases - Mandatory search-first workflow: Resolves entity names and property names to confirmed Q/P IDs with the wikidata_search tool before any SPARQL is written. - Domain-specific property sets: Ships reference files for cities, persons, organisations, creative works, and chemical compounds, each with the correct P31 types, recommended VALUES property lists, and known pitfalls. - Ready-made SPARQL patterns: Includes label lookup, substring search, class instance listing, multi-hop traversal, and qualifier-based full statement queries with auto-injected prefixes. - Use Case: Ask "When was Marie Curie born and what were her occupations?" and the Skill searches for her Q-number, loads the person domain file, and runs a query using P569 and P106 to return verified facts. ## Quick Start Ask the assistant to look up a real-world fact in Wikidata, for example: "Use Wikidata to find the population and country of Munich."
