Specifies that untrusted retrieved or user content cannot authorize tools or exfiltrate secrets, that the model follows a documented instruction hierarchy, and that outputs are filtered before privileged side effects.
Published Sep 12, 2026
No related Blocks have been published yet.