injectcheck

Scan what AI agents read for prompt injection: hidden unicode, comment instructions, exfil, lures.