I want to examine secrets inside an AX Task separately. Even if a secret is injected through a safer mechanism, the agent can obviously read its own environment.

on Bluesky