Back to home
Legal

How we clean your data

Cleaning runs on our servers before data is stored. The nine categories below are removed and replaced with a marker such as [EMAIL_REDACTED].

Beta · subject to change

Nine categories that are removed

CategoryCoverage
Seed phraseBIP-39 12/24-word phrases with seed/mnemonic/recovery markers
Private keyPEM and OpenSSH blocks
WalletEVM 0x…, Bitcoin bc1/1/3
API keyOpenAI sk-, Stripe sk_live_, AWS AKIA, Google AIza, GitHub ghp_
Credit card13–19 digits, Luhn-validated so ordinary numbers aren’t mistakenly caught
Identity numberKorean RRN, 6–7 digits
EmailEmail address
PhoneE.164, Korean format, and common patterns
URLhttp/https and www

Cleaning order

The order is deliberate: seed phrase, private key, wallet, API key, credit card, identity number, email, phone, then URL. More specific categories are processed first so that broader patterns like URL don’t swallow part of a sensitive value and leave a fragment behind.

What you should know

Cleaning is pattern-based, not perfect. Personal data in an unusual format can slip through. Review the preview before approving, and don’t upload documents that truly must not leave your device.

The original file is stored separately from the cleaned version and is not shared. You can revoke consent at any time; the history is append-only, so every grant and revocation stays readable. The ConsentRegistry contract already exists on Robinhood Chain for on-chain recording, but it isn’t used automatically yet, so consent records are currently stored in our database.

You can also delete your own archive from Archive: the original file and its cleaned result are permanently deleted.