Prompt Injection Firewalls at Actions

Why this idea is here

What the evidence establishes.

Source-backed premise: the cited captures establish agent identity and security-governance pressure. Editorial projection: The strongest defense will inspect proposed actions and authority, not merely filter model inputs and outputs. The sources do not prove that outcome; the observable adoption test is whether security evaluations prevent harmful actions by checking authority and proposed effects at runtime.