The tool combines Jev judgments with local checks, but masking only hides its output after the text reaches TypeSafe. It can mistake public figures for private names or miss details split across chunks, so its findings still need review.
How you can use it
Start with made-up text containing names and numbers you expect the checker to find. A developer can compare its warnings with your list before choosing whether serious warnings should make a scan fail. Review mistakes yourself. The text being scanned goes to TypeSafe, an outside AI service; hiding names and numbers in the results happens afterward, not before that text leaves your computer.