Log Anonymizer

About the Log Anonymizer

Sharing a log is how most support requests, bug reports and incident write-ups begin. It is also how customer email addresses, internal IP ranges, session tokens and occasionally card numbers end up in a public issue tracker, where search engines find them.

This tool anonymises a log before it leaves your machine. It recognises email addresses, IPv4 and IPv6 addresses, phone numbers, UUIDs, MAC addresses, API keys and JWTs, card numbers, and usernames embedded in file paths, and replaces each with a placeholder.

The replacement style is where the design earns its keep. Redacting everything to a single marker destroys the log's meaning — you can no longer see that three failed requests came from one address. The default here assigns numbered tokens instead, so every occurrence of a particular value becomes the same tag and different values get different tags. The pattern of the incident survives while the identities do not, which is what makes the log still worth reading.

Card numbers are checked against the Luhn algorithm before being masked, so a sixteen-digit build identifier or order reference passes through untouched while a real card number does not. The JWT and API-key patterns look for the structural shapes those take, and for the long opaque values that follow words like bearer, token and secret.

Say the honest thing about what this can promise: pattern matching finds the shapes it knows. A customer's name in a free-text field, an internal hostname that reveals your architecture, a URL that identifies an account — none of those look like anything in particular, and none will be caught. Run the tool, then read the result before you post it. The tool removes the categories of secret that are mechanically recognisable; the judgement remains yours.

How to use it

1. 1Paste the log into the input. Large files are fine — it is all processed locally.
2. 2Choose the replacement style. Numbered tokens keep the correlation between repeated values, which is usually what makes a log worth reading.
3. 3Switch individual categories on or off. Masking usernames from file paths is off by default because it often obscures the very path you are asking about.
4. 4Optionally filter to lines containing a keyword, so you share the relevant portion rather than the whole file.
5. 5Read the result before you post it. Pattern matching cannot find a secret that does not look like one.

Questions

Is my log uploaded to be processed?

No, which is the reason this tool exists in this form. A log is exactly the kind of file you should not hand to a third party, so the matching and replacement run in your browser and nothing is transmitted.

Why are some sixteen-digit numbers masked and others not?

Card numbers are validated with the Luhn checksum first. Real card numbers satisfy it; sequential build numbers and order references almost never do. That check is what stops the tool from mangling every long number in the file.

What does the numbered style actually do?

It gives each distinct value a stable tag, so one address becomes the same tag everywhere it appears and a second address gets a different one. You keep the ability to see that several events shared a source without knowing what that source was.

Is the output safe to publish?

Safer, not safe. Everything with a recognisable shape has gone. Names, hostnames, account identifiers in URLs and anything else that reads like ordinary text will still be there, so the result needs a human read before it goes anywhere public.