Agent-runtime security layer that detects prompt injection and destructive actions in webpages, emails and files, with span-level redaction and tool-call controls
No reviews yet. Be the first to review Unplug AI.