Keep the warehouse trustworthy: dbt tests (not_null/unique/accepted_values/relationships) gating the build in CI, source-freshness checks, model contracts at consumer boundaries, singular tests for business invariants, and anomaly detection beyond schema tests.
Installs into .claude/skills of the current project.
Are you the author of Data Quality Testing?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/mcorbett51090-data-quality-testing)
---
name: data-quality-testing
description: "Keep the warehouse trustworthy: dbt tests (not_null/unique/accepted_values/relationships) gating the build in CI, source-freshness checks, model contracts at consumer boundaries, singular tests for business invariants, and anomaly detection beyond schema tests."
---
# Data Quality Testing
## Test like code
not_null / unique / accepted_values / relationships on the columns that matter; **build fails** on violation. Untested transforms = silent downstream corruption.
## Freshness
Assert source freshness; **stop** rather than serve stale marts as fresh.
## Contracts
Model contracts (names/types/constraints) on consumer-facing models so upstream changes can't silently break a dashboard.
## Beyond schema
Singular tests for business invariants (line items sum to order total). **Anomaly detection** (row-count drop, distribution shift) catches what schema tests miss -> significance to `applied-statistics`.
