Data contracts and tests as code. Schema enforcement, freshness / completeness / uniqueness / range checks, anomaly detection on key metrics, alerting on data-SLO breaches. Treat data like product — versioned, tested, with explicit consumers + producers + SLOs. Distinct from `testing-strategy` (which tests application code); this skill tests data correctness continuously as it flows through pipelines and warehouses. Data Engineer owns this; analysts consume the contracts; alerts route through incident-runbook. Use whenever new tables are being added, when data quality issues surface in dashboards, or when establishing data-SLOs.
