Prompt Injection Check Screens Text Before It Reaches an AI

Why it surfaced

This is a narrowly focused tool with a clear operating rule: treat pasted instructions as untrusted input. It flags things such as instruction overrides, role hijacking, prompt extraction, hidden content, obfuscation, and selected encoded attacks—while plainly admitting that a warning is not a safety guarantee.

A browser-side scanner that checks pasted emails, documents, webpages, and prompts for common prompt-injection warning signs.