A spreadsheet can look ready and still contain blank cells, duplicate rows, or inconsistent formats. The open-source Data Quality Skill documents a read-only screening workflow you can run before asking an agent for conclusions.
Define the scope
Record the file, sheet, and range you checked. If you inspected one tab, keep your conclusions within that tab.
Confirm the rules
A column that usually has a value does not automatically require one. Separate confirmed business rules from candidate rules inferred from the data, then check missing values, duplicates, and formats.
Name what the check missed
A passing result does not establish that the data is accurate or current. Keep the original file and list the checks you skipped. For XLSX, the script does not recalculate formulas, so cached values may be stale.
Try asking your agent for a read-only review that starts with scope, asks you to confirm candidate rules, and ends with results and unchecked items.
Data Quality Skill source and MIT license · Skillπ AI skill marketplace
Top comments (0)