What problem does it solve? Writing repository and query service tests in Go often drifts into mocks, partial assertions, or ad-hoc fixtures that fail to verify what actually changed in the database. This Skill enforces a convention where every BDD scenario from the data-model document is replayed on a real PostgreSQL instance, with all tables compared row-by-row before and after each operation. ## Core Features & Use Cases - Full-table BDD verification: Repository tests drive restore → real aggregate operation → save, then assert every table's rows match the document's After state, including DB constraint translation, optimistic-lock conflicts, concurrency, and NotFound cases. - Query service testing: Seed Before rows, call the read method, and compare returned DTOs across filter, paging, NULL, and empty-result perspectives. - Deterministic infrastructure: A single dockertest-based rdbtest.Start in TestMain, typed Seed{Table}/Read{Table} helpers generated via sqlc, and a check-bdd-coverage.py script that verifies every BDD ID appears either as a test case or a documented omission. - Use Case: Given a data-model document with BDD-001 through BDD-010 for a room reservation aggregate, generate TestReservationRepository_ApplyHeld and sibling tests that seed rows, run concurrent transactions, and assert all eight tables match the documented After state. ## Quick Start Ask the AI to write repository tests from this data-model document's BDD scenarios, verifying all tables on a real PostgreSQL database.
