Csv Data Cleaner

Raw CSVs are almost never analysis-ready: inconsistent nulls (N/A, none, -), stray whitespace, duplicated rows, mixed formats. This skill profiles any CSV with the included Python script (standard library only — runs anywhere Python 3 runs), then applies explicit cleaning rules and produces a markdown audit report that logs every row it touched. Nothing is ever dropped silently.

Give you a repeatable, auditable two-pass workflow for dirty CSVs: profile first, choose cleaning rules deliberately, apply, then verify with a second audit run. The script detects the delimiter, profiles every column (null counts with the distinct null tokens found, inferred types, duplicate rows, min/max for numerics), and in clean mode emits a cleaned CSV plus a row-by-row change log.

• A CSV import, scrape, or export has messy values before analysis or a database load.

• You need to deduplicate rows, normalize null tokens, or trim whitespace safely.

• A colleague asks "what did you change in this file?" and you want a written answer.

• The file isn't CSV-like at all (Excel, JSON, XML) — use a parser for that format.

• Cleaning requires domain judgment per row (e.g. which of two conflicting records is true) — the script only handles mechanical rules; judgment calls stay human.

• Sensitive data you aren't authorized to read — the profiler reads the whole file.

• You need date-format normalization or currency conversion — out of scope by design (see Limitations).