Inbox Injection Guard
Screens your email for hidden instructions aimed at AI assistants, like white text, zero-width characters, and 'ignore previous instructions' lines.
Problem
When an AI assistant reads your inbox, a malicious email can hide instructions in white text or invisible characters that try to hijack it.
Solution
It scans incoming email for hidden prompt-injection tricks such as white text, zero-width characters, and 'ignore previous instructions' lines, and flags them before your assistant acts on that mail.