Commands let you act on a device without physically touching it or setting up remote desktop access. Every command you run is logged, with its full output, against the device it ran on.
What each command does
| Command | What it does |
|---|
| Restart | Restarts the device |
| Shutdown | Powers the device off |
| Refresh Inventory | Forces an immediate software/hardware scan, instead of waiting for the next scheduled one |
| Collect Logs | Pulls the agent’s own log file back to your dashboard — useful for diagnosing a misbehaving agent |
| Run Script | Executes an arbitrary shell command or script on the device |
| Update Agent | Pushes a specific agent version to the device (see Installing your first agent for how updates work) |
Running one
From the Commands page (New Command) or from any device’s own page — running from the device page skips having to pick a target, since you’re already looking at it.
Picking a target follows the same shape everywhere in AssetGullak: a single device, a whole location or department, or your entire company. Quick actions are listed first, Run Custom Script is separate and secondary — you shouldn’t have to scroll past a raw script box to restart a device.
Run Script is more restricted, on purpose
Every other command does one specific, bounded thing. A script can do anything a person sitting at that machine could do — which is why it’s gated differently:
- Only Owner, Admin, and IT Admin can run scripts — IT Operator can issue every other command type but not this one. See Roles & Permissions for why.
- Scripts run with a 5-minute timeout — if it hasn’t finished by then, it’s killed and reported as timed out, so a hung script can’t tie up a device indefinitely.
- Output is capped at roughly 20,000 characters — a script that produces a huge amount of output gets truncated rather than overwhelming the result view.
Command results
Every command’s detail view shows exit code, stdout, and stderr. When you target more than one device, you get a batch — one job with a per-device breakdown, so you can see at a glance which devices succeeded and which didn’t, rather than opening each one individually.
Cancelling a batch stops anything still pending — devices that already claimed and started running their command finish normally; only the not-yet-started ones are cancelled.
Where results are logged
Every command run appears on the target device’s own Timeline tab, and the full batch history lives on the Commands page — searchable by status (pending, running, completed, failed). 
