Hi @todd ,
@Tomer_Noy 's answer is the definitive one here — it's by design, and worth understanding why, because that rationale is also your answer to the auditor: anyone who can manage administrators and permission profiles can create a profile and account for themselves with any rights. So a "Manage Administrators without Install Policy" profile would give the appearance of separation without the substance. That's a stronger audit position than a control that looks separated but isn't.
Framed for an audit conversation, you have three layers to work with:
1. Preventive (technical) — the SmartTask approach Tomer described. Two useful details when you build it: the Web Request action posts the trigger data as JSON to your endpoint, and for a "Before" trigger your endpoint replies HTTP 200 with a JSON body containing result (and optionally message) to allow or abort the operation. So you can genuinely block a publish/install, and return a message the admin sees.
Practical safety, on top of Tomer's warning: test it against a narrow scope first, keep a documented break-glass superuser account, and make sure your endpoint fails open or closed deliberately — decide in advance what should happen if the webhook is unreachable, because that's the scenario that bites people.
2. Detective (accountability) — every administrator action is recorded in the management audit logs (Logs & Monitor). For many audit frameworks, when preventive segregation isn't architecturally available, a documented detective control — audit logging, alerting on policy installs by the identity-management admin, and periodic review — is an accepted compensating control. Pair that with the SmartTask and you have both layers.
3. Process/architecture — the option worth considering if the auditor wants true separation of duties: take administrator provisioning out of interactive SmartConsole entirely and drive it through the Management API from your IAM/PAM platform. The privileged credential lives in the PAM vault with approval workflow, just-in-time elevation and session recording, so the segregation of duties is enforced in the process layer where auditors usually expect to see it. To be clear, this doesn't bypass the design — the API identity still holds equivalent permissions — but it moves the control to an auditable, approval-gated workflow instead of a standing human privilege. That distinction is usually what satisfies the finding.
Also worth having in place regardless: 2FA for administrators and trusted-client/login restrictions on the management.
Hope this helps!
Best regards,
Jorge Dias Junior