dataset-generatelisted

# dataset-generate — synthetic, business-coherent test data `qaia-core` only ever produces small inline examples inside a scenario — test data is a separate concern, deliberately kept out of the generator so a scenario stays readable and the data stays reusable. This skill is the separate producer of **standalone, reusable datasets**: entity collections and scenario-oriented "cases" rich and coherent enough that `qaia-playwright:automate`'s generated tests can seed real state from them instead of inventing a literal per test. ## Prerequisites - The source US (`.qaia/state/<US-ID>/00-source.md` / `01-extraction.md`, or an equivalent gold-set-style file) — the acceptance criteria are what the dataset's entities, fields and boundaries are derived from. - Ideally the design/test book (`.qaia/state/<US-ID>/03-design.md`, `.feature` files under `.qaia/testbooks/<US-ID>/`) so cases can cite the actual derived condition/scenario IDs. If only the US is available, generate from its acceptance criteria alone and say so — do not block on a test book that does not exist yet. ## Steps 1. **Model the entities first.** Read the acceptance criteria and extract the nouns and relationships they imply (reference records, actors, transactional records, the foreign keys between them). Write the entity list and its relationships down before generating a single row — a dataset invented row-by-row with no schema is how referential integrity breaks silently. 2. **Business-co