Data Quality Auditor
Run deterministic checks from a reviewed rule file:
python scripts/validate_dataset.py data.csv --rules quality-rules.json
Supported rule keys are required_columns, non_null, unique, types,
ranges, allowed_values, and row_count. The report contains counts and row
numbers but never raw failing values. The command exits nonzero on violations;
use --warn-only only when a non-blocking audit is explicitly intended.
Rule example
{
"required_columns": ["id", "created_at"],
"non_null": ["id"],
"unique": ["id"],
"types": {"id": "integer", "created_at": "datetime"},
"ranges": {"score": {"min": 0, "max": 1}},
"allowed_values": {"status": ["active", "inactive"]},
"row_count": {"min": 1}
}
Review the rules and dataset authorization before running. A passing report proves
only the declared checks; it does not establish correctness, fairness, fitness for
purpose, or permission to use the data.
1---2name: data-quality-auditor-23domain: data4skill_type: skill5description: >-6 Validate CSV, TSV, JSON, or JSON Lines data against an explicit JSON rule set and7 emit a redacted, machine-readable pass/fail report. Use when checking required8 columns, nullability, uniqueness, types, ranges, allowed values, or row-count9 expectations before analysis or ingestion. Do not use to repair data or infer10 unstated business rules.11license: MIT12tags: [data, quality, validation, schema, rules, csv, json]13metadata:14 source: https://skillmd.com/skills/knuckles-team/data-quality-auditor-215 version: '1.6.0'16 author: Genius17---1819# Data Quality Auditor2021Run deterministic checks from a reviewed rule file:2223```bash24python scripts/validate_dataset.py data.csv --rules quality-rules.json25```2627Supported rule keys are `required_columns`, `non_null`, `unique`, `types`,28`ranges`, `allowed_values`, and `row_count`. The report contains counts and row29numbers but never raw failing values. The command exits nonzero on violations;30use `--warn-only` only when a non-blocking audit is explicitly intended.3132## Rule example3334```json35{36 "required_columns": ["id", "created_at"],37 "non_null": ["id"],38 "unique": ["id"],39 "types": {"id": "integer", "created_at": "datetime"},40 "ranges": {"score": {"min": 0, "max": 1}},41 "allowed_values": {"status": ["active", "inactive"]},42 "row_count": {"min": 1}43}44```4546Review the rules and dataset authorization before running. A passing report proves47only the declared checks; it does not establish correctness, fairness, fitness for48purpose, or permission to use the data.