# AI Data Quality Monitoring Rules

Companion to **AI Agents for eCommerce Data Quality Monitoring**.

Deterministic rules **detect**. The model **investigates** (optional). Do not let the model be the validator for required fields.

Catalog **publish** gates remain in [`catalog-validation-rules.md`](catalog-validation-rules.md).

## Monitor catalog (examples — replace thresholds)

| Check | Deterministic rule | Agent may | Agent must not |
| ----- | ------------------ | --------- | -------------- |
| Missing required attrs | `required_attrs` empty on live SKU | Draft enrichment | Auto-publish |
| Duplicate customers | Same email + different `customer_id` | Flag merge candidate | Merge accounts |
| Incorrect prices | Storefront price ≠ price service `asOf` | Open ticket | Execute live price |
| Inventory anomaly | ATP jump > N% with no receipt | Investigate receipts | Adjust ATP |
| Sync lag | `asOf` older than SLA | Page ops | Invent stock |
| Suspicious change | Price/cost change outside window + no PO | Risk flag | Accuse a person |

## Output schema (one finding)

```json
{
  "finding_id": "DQ-DEMO-001",
  "domain": "products",
  "rule_id": "required_attrs_empty",
  "severity": "high",
  "evidence_tool": "scanCatalogCompleteness",
  "evidence_ref": "sku:DEMO-SKU",
  "recommended_action": "draft_enrichment",
  "human_required": true
}
```

Demo ids are fixtures. Do not treat counts in a blog post as your backlog.

## Related

- Catalog validation (produce quality): [`catalog-validation-rules.md`](catalog-validation-rules.md)
- Data layer: [`ai-ready-business-data-layer.md`](ai-ready-business-data-layer.md)
- Exception pattern: [`exception-agent-pattern.md`](exception-agent-pattern.md)