Codex Six-Dimension CSV Data Quality Audit
Audit completeness, accuracy, consistency, timeliness, validity, and uniqueness with an issue and remediation log.
中文Install in Codex
Prepare a local sample, data rules, update timestamps, and a reference that can substantiate accuracy.
- Copy the pinned `skills/codexkit-data-quality-auditor/` directory into `.agents/skills/codexkit-data-quality-auditor/` in the project.
- Have Codex read the sample and rules, then specify the denominator and threshold for each check.
Example task
Use ten synthetic `catalog.csv` records, including one missing price, two rows sharing a SKU, and a negative price. Supply an allowed price range and a reference SKU table.What to check
The report gives counts for missing prices, duplicate groups, and negative values. Accuracy and timeliness ratings use the reference data actually supplied.
Original SKILL.md on GitHub · Updated
Frequently asked question
How can I assess accuracy without a reference?
Record that accuracy evidence is insufficient, and identify the reference source or sampling needed for validation.