anonymise.py · client-side scrubber
Anonymise your export before you send it.
A single Python file, no dependencies you don't already have. Your admin runs it on the export; the output is what you upload.
Request the script by email. We are finalising the packaged version — write to viktor@arctura.eu with the subject "anonymise.py" and we send the current build back the same working day, along with a one-page checksum note so your security team can verify it before running.
What it does
- Reads any CSV or JSON export from Zendesk, Intercom, LiveChat, Freshdesk — or any other helpdesk that produces one row/object per message.
- Detects and removes emails, phone numbers, IBANs, wallet addresses, card and account numbers, document numbers, dates of birth.
- Replaces player and agent names and identifiers with stable hashes so repeat-contact counting still works.
- Rounds exact amounts to ranges (
10–50 EUR,50–100 EUR, and so on). - Does not touch columns it doesn't recognise — it appends anonymised versions rather than overwriting, so nothing breaks if your export has an unexpected shape.
- Writes a companion
.logfile listing what was scrubbed and how many hits per rule, so you can spot-check the result.
What it needs
- Python 3.9 or later. Standard library only. No pip install.
- Read access to your export file and write access to a folder you choose.
- About 30 seconds to run per 10,000 messages on a laptop.
Alternative: we run it together
If you'd rather not run a script on your machine, we screen-share for 15 minutes on your admin's laptop and run it together. Nothing gets copied to us during that call — the anonymised output stays on your side until you're ready to upload.