What problem does it solve? AI-generated or human-written BDD tests can be green while verifying nothing: empty oracles, self-blessing goldens, and non-discriminating test data slip into suites and erode trust. This Skill performs a structured expert review of pytest-bdd/Gherkin PRs in the qubership-envgene repository to catch those defects before the suite is trusted. ## Core Features & Use Cases - Four-axis validation: Checks scenario completeness against docs and code, scenario validity (doc conformance, oracle strength, oracle independence, determinism), test-data completeness, and test-data validity. - Doc-vs-code divergence resolution: Builds a numbered divergence list, determines which side violates the authoritative contract, and drives resolution with the user via issues, docs PRs, or @xfail(strict) markers. - Canonical verdict report: Produces a per-scenario verdict table (valid, weak, invalid, missing, not needed), proposed Gherkin drafts, a test-data matrix, and notes, published as a PR comment only on explicit approval. - Use Case: A PR adds Gherkin scenarios for the effective-set generator. Invoke /bdd-test-review to verify each scenario exercises the real CLI, that payloads fail at the claimed point, and that goldens are independently derivable from the documented contract. ## Quick Start Ask the assistant to run /bdd-test-review on a specific pull request number that adds or changes BDD scenarios in the qubership-envgene repository.
