data-quality-operations_skill

Overview

This skill provides data quality validation patterns and daily operational commands for routine freshness, completeness, and anomaly follow-up. It standardizes inputs, owner assignment, and artifact capture to make daily checks repeatable and auditable. The package includes profiling, rule validation, anomaly tracking, and workflow templates for checklists and reports.

How this skill works

The skill runs a sequence of commands to profile a dataset, validate it against predefined rule sets, and open or track anomalies for investigation. It enforces collecting target, impact, owner, and deadline before action, and saves outputs and timestamps for audit and handoff. A structured report template is included for concise post-task summaries.

When to use it

• Daily or nightly freshness and completeness checks for production datasets
• When a scheduled alert indicates a potential anomaly in metrics
• Before handoffs to downstream teams or after incident triage
• During audits that require repeatable validation evidence
• When onboarding a new dataset and establishing baseline checks

Best practices

• Always assign a single explicit owner and a clear deadline before starting validation
• Keep timeline notes concise, timestamped, and linked to the saved artifacts
• Use rule sets for repeatable validation and profile datasets regularly to detect drift
• Open anomalies with context (metric, impact, urgency) rather than ad hoc notes
• Save outputs and reports in a consistent location for auditability and handoff

Example use cases

• Run dq profile --dataset sales_db.daily to capture current column distributions and null rates
• Execute dq validate --rule-set 42 nightly to confirm freshness and row-count thresholds
• Open an investigation with dq anomaly --open --metric order_processing_time when SLA breaches occur
• Use workflow checklist --from templates/checklist.md during a dataset ownership transfer
• Generate a post-mortem summary with workflow report --from templates/report.md after resolving an anomaly