PII Redactor & Text Scrubber

Detectors

Notes on each detector

• Email uses a pragmatic RFC 5322 subset — matches almost every real address without matching every plausible one.
• Phone covers E.164 (+1 415 555 0123), US formatted ((415) 555-0123), and other 7–15 digit groups with common separators. Ambiguous digit runs adjacent to code (like 0x1234abcd) are ignored.
• Credit card catches 13–19 digit strings that pass the Luhn checksum. Sequences like 1234 5678 9012 3456 that fail Luhn are left alone to avoid false positives.
• SSN matches the US AAA-GG-SSSS format and rejects known invalid blocks (000, 666, 900-999 area, 00 group, 0000 serial).
• IPv4 validates each octet 0–255 and skips version-like triplets embedded in longer numbers.
• IPv6 covers full 8-group and ::-compressed forms, including embedded IPv4 like ::ffff:1.2.3.4.
• IBAN validates the mod-97 checksum after the country and check digits, so DE89 3704 0044 0532 0130 00 matches but a random string of the same shape won't.
• UUID matches versions 1–5, with or without hyphens.
• URL and MAC are off by default because they often aren't PII — toggle on when reviewing logs or configs.

In consistent token mode the same value maps to the same tag every time it appears — useful when you need the scrubbed text to remain internally referential (e.g. so a single user's actions still cluster in a shared log). The hash mode does the same deterministically across sessions when you use the same salt.