Find the API keys and personal data in your dataset before you upload it.
Drag in a .jsonl, .csv or .parquet — gigabytes are
fine — and Scrubjay streams it through
26 recognisers in your browser. You get back a
pseudonymised copy that still trains, a findings report, and an encrypted map only you can
reverse.
1 Choose a dataset
Drag a dataset here
.jsonl, .jsonl.gz, .json, .csv,
.txt or .parquet. Nothing is uploaded.
2 Before it runs
Recognisers — every one publishes what it does not match
3 Scanning
Keep this tab in front. If your phone locks the screen, the browser pauses the scan.
4 What it found
5 Your artefacts
The mapping file
mapping.enc reverses every pseudonym. It is sealed with AES-GCM-256 under a
key derived from this passphrase at 600,000 PBKDF2 rounds. The passphrase is
not stored anywhere — if you lose it the mapping is gone.
Copying puts it on the system clipboard, which macOS and Windows sync between devices.
Your project key
Save this if you will scrub another shard of the same dataset. The same key produces the same pseudonyms, so the shards stay joinable. Anyone holding it can match a candidate list against every pseudonym instantly — treat it as a secret.
Reverse a mapping
Drop a mapping.enc you made earlier to look a pseudonym back up. It is opened
in this tab and nothing is sent anywhere.