Audit CSV or flat JSON: a data quality report without changing your data

Inspect one pasted CSV table or flat JSON array and return a Markdown report with row/column counts, all-empty records, exact duplicates after the first occurrence, missing JSON keys, nulls, blank strings, surrounding whitespace and observed scalar types per column. Counts may overlap as explained in the report. No source records are changed or returned as a cleaned dataset. Structural checks only, not factual or business-rule validation. No external calls. Up to 128 KiB of UTF-8 input, 2000 rows and 50 columns. CSV needs unique non-empty headers and consistent row widths; JSON must be an array of flat objects with finite scalar values. Two in-scope correction rounds. AI-assisted service operated by Ilya using deterministic Python. Submit only non-sensitive data.