AI Autonomy — The Playbook Library

A playbook is a set of script steps that runs on a device through the Plenix agent. AI Autonomy comes with 159 built-in playbooks, and you can add your own.

Before you begin

• AI Autonomy is part of the Enterprise plan.
• A playbook can only run on a device that has the Plenix agent installed and online.

Find a playbook

Open AI Autonomy and stay on the Playbook library tab.

• Search by name, description, tag or key.
• Click a category to narrow the list. There are 12 categories: Health & Performance, Storage & Cleanup, Network, Security, Updates & Patching, Services & Processes, Printing, Apps & Office, Hardware & Inventory, Users & Identity, Servers, and Power & Restart.
• Filter by operating system, risk, mode, or Built-in only / My playbooks.

The numbers across the top of the page are clickable. For example, Fully automatic lists only the playbooks that run without approval, and Failed opens the runs that failed.

Risk tiers

• Read-only. Collects information and changes nothing.
• Low risk. Small, safe changes, such as clearing temp files.
• Medium risk. Changes the user may notice, such as restarting a service.
• High risk. Disruptive actions, such as a reboot.

Run a playbook on a device

1. Click a playbook to open it.
2. Under Run on a device, search for the device and pick it.
3. Optionally, enter a ticket ID to link the run to a ticket.
4. Click Run playbook.

What happens next depends on the playbook's mode. In Fully automatic mode it runs straight away. In Suggest only or Requires approval mode it waits in Approvals. In Disabled mode it is refused.

A playbook only runs on its own operating system. If you pick a Linux device for a Windows playbook, Plenix tells you and nothing runs.

See what a playbook does

The playbook window lists every step with its full script and timeout, and the playbook's recent runs. Click a run to open its transcript, which shows the output of each step.

Copy a playbook

Click Copy on any playbook, built-in or your own. The copy is yours to change: it opens in the editor and records which playbook it came from. Built-in playbooks cannot be edited or deleted, so copying is how you make your own version of one.

Write your own playbook

1. Click New playbook.
2. Give it a name and description, and choose a category, operating system and risk tier.
3. Add up to 20 steps. Each step has a name, a script and a timeout between 5 and 600 seconds. Windows steps run as PowerShell; Linux and macOS steps run as shell scripts.
4. Use the arrows to change the order of the steps.
5. Click Create playbook.

Steps run in order, and the playbook stops at the first step that fails.

Only mark a playbook Read-only if its scripts change nothing. Read-only playbooks run automatically by default.

Edit or delete your playbooks

Your own playbooks carry a Custom label and have Edit and Delete buttons. Deleting a playbook does not delete its past runs.

When a run of your playbook goes to approval, it records the steps as they were at that moment. The approver approves exactly what will run, even if someone edits the playbook in the meantime.

The Runs tab

The Runs tab lists recent runs with their status, device, who started them and the mode they ran under. Filter by status, and click a run to open its transcript. Runs waiting for a decision link to Approvals.

Related guides

• Configure Autonomy Policies
• Approve or Reject Pending Actions
• Use AI Ticket Triage