Home

CSV Anonymizer Wiki

A static browser page that fuzzes and redacts CSV data. Nothing is uploaded: the file is read and processed by JavaScript in your own tab.

Use it online, or download the repository and open index.html locally for an offline copy.

Pages

• What it is, and what it is not: read this before using the output for anything that matters
• How columns are detected: the type of each column is guessed from its header, and the guess is often wrong in ways you can predict
• Choosing settings: presets, redaction against fuzzing, and which to use when
• Practical limits: what breaks in the output, and what the browser can handle

The short version

This is a fuzzer. It produces a file that looks like your data, keeps the same shape, and is useful for demos, screenshots, bug reports and test fixtures.

It is not anonymisation in the sense a regulator or a privacy engineer means. It does not provide k-anonymity, l-diversity or differential privacy, and it does not stop re-identification through the combination of columns it left intact. The README's suggestion to "confidently share" the output is fine for a test dataset and wrong for a dataset about real people.

If the answer to "what happens if someone re-identifies a row" is anything other than "nothing", see What it is, and what it is not first.

Where it is genuinely good

• The data never leaves your machine. No upload, no server, no account. For a sensitive file this is worth more than most of the processing that follows it
• The structure survives. Columns, row count and formats stay usable, so the output drops into the same pipeline as the original
• It runs offline. Downloading the repository and opening index.html gives you a copy that works on an air-gapped machine

Licence: AGPL-3.0, which is strong copyleft rather than permissive: reusing this code in a networked service carries obligations. The README describes it as permissive, which is a mistake in the README, not in the licence.