Skip to main content
Working With Your Agents cover — your terminal, unchanged. Your organization has put AI coding agents under AgenShield. This page is what you actually need to know. The short version: almost nothing changes.

​
What changed

Everything your agents run, read, and connect to is now evaluated against your organization’s policy, and recorded. Practically: There is nothing to enable per agent: every agent AgenShield detects is governed by the policy your organization publishes. You do not need a new command to launch anything, and nothing about your day-to-day workflow changes.

​
Checking status

The quickest answer is the menubar icon; the dashboard shows the detail. From the terminal:
The bottom line summarizes the machine:

​
When something is blocked

A blocked action fails the way an ordinary permissions problem fails. There is no AgenShield dialog in your terminal: Do not work around it. Every block is recorded with the rule that caused it, and that record is exactly what your administrator needs to fix the policy. Send them:
  • what you were trying to do
  • roughly when it happened
  • the error your agent reported
Being in monitor mode does not guarantee nothing is blocked: individual rules can be promoted to enforce and still block an action (see the three enforcement modes). If your administrator confirms no rule recorded a block, the failure has another cause — check Common issues.

​
Signing in

Open the AgenShield menubar icon and click Log in — it opens your organization’s sign-in page in the browser. The same button is on the dashboard. Signing in links this Mac to your user account so policy can apply rules scoped to your team, role, or group; until you do, only device-wide rules apply. Prefer the terminal? agenshield login starts the same browser sign-in.

​
Everyday commands

Full list: CLI reference.

​
Frequently asked

Can I turn it off? No — policy is managed centrally by design, and there is no local override. Ask whoever administers AgenShield for your organization. Is it watching what I do? No. Enforcement and recording are scoped to AI agents. Your own commands, files, browsing, and other applications are not observed. See Privacy and data handling. Will it slow my Mac down? Decisions are made in the kernel and cached. You should not notice it in normal use. If you do, that is a bug — run agenshield doctor and report it. Can it lock me out of my Mac? No. The processes and system paths macOS needs can never be denied — those limits are compiled into the signed security extension and are not configurable. If AgenShield ever starts denying an unusual volume of ordinary activity, it stands itself down automatically and heals itself. And if the host account is ever blocked anyway, an administrator can disable enforcement locally in seconds — see It cannot lock you out.