quality-audit

name	quality-audit
description	Run a full data-quality audit on a table or pipeline: quality score, rule checks, anomalies with context, and concrete fixes ranked by impact.
user-invocable	true
argument-hint	[table or pipeline, e.g. analytics.orders]

Quality Audit

Audit a dataset's health end to end. The user names a target via "$ARGUMENTS".

Examples

• /data-workers:quality-audit analytics.orders
• /data-workers:quality-audit did anything drift in staging.customers this week?

Step 1 — Baseline

Call mcp__data-workers__get_quality_score for the current weighted score, then mcp__data-workers__get_quality_summary for the dimension breakdown (completeness, freshness, validity, consistency, uniqueness).

Step 2 — Check and explain

• mcp__data-workers__run_quality_check to execute the active rules against the target.
• mcp__data-workers__get_anomalies for statistical outliers, then mcp__data-workers__get_anomaly_context on each significant anomaly so the report explains why it fired, not just that it fired.

Step 3 — Recommend

Report: score with trend, failed rules with affected row counts, anomalies with context, then 2-3 concrete next steps ranked by impact. If the target is a pipeline output, offer mcp__data-workers__create_quality_tests_for_pipeline to codify the missing checks — ask before creating anything.

Guardrails

• Creating tests or SLAs changes state — always confirm with the user first.
• Never invent scores or row counts; report only what the tools return.
• On first run without warehouse credentials the server uses in-memory sample data — say so when reporting.