Paste text, redact common identifiers, then copy the safe version for your LLM prompt.
This runs entirely in your browser.
No ReactSingle HTML file
Tip: add proper nouns you know are identifying.
Input
Output (redacted)
What this will (and won’t) catch
Good at: emails, URLs, IPs, common phone patterns, UK postcodes, and any custom terms you add.
Not reliable for: people’s names, organisations, job titles, or “identifiability” in general — add those via custom terms.
“Card-like numbers” is conservative and may redact non-card numbers that look similar.
If you need stronger guarantees, use a dedicated PII detection library/service (NER-based), but that usually means sending data somewhere or bundling a heavier model.