npm package · live · 2026-10
part of foxmate
foxshield
find hidden text and prompt injection before an agent reads a page
the problem
a web page can hide text that tells an ai agent what to do. the agent reads it and the person never sees it.
what i built
foxshield scans a page for hidden text and prompt injection. its sanitize step drops the hidden text and marks flagged text as untrusted data. a CLI fails a build when a page carries hidden instructions.