Single shortcut, anywhere
Ctrl+K opens the palette from any page, as does a click on the Search... box in the header. ↑ ↓ to move, Enter to execute, Esc to close.
Keyboard-driven access to modules, navigation and Proxmox VE operations, without leaving your current page. The palette is always available. Some result types and commands appear only when a specific cluster is selected.

Ctrl+K opens the palette from any page, as does a click on the Search... box in the header. ↑ ↓ to move, Enter to execute, Esc to close.
One textbox for modules, QEMU VMs, nodes and commands. Plain text searches the modules. The prefixes (vm:, node:, ip:, >) switch to guests, nodes and commands.
start, stop, restart, console, create snapshot: operate on any VM/CT without clicking through the resource grid.
Results are filtered by the cv4pve-admin permissions of the current user. Commands the user can't execute don't appear.
Commands that need input open a typed parameter dialog (VM picker, text, boolean): no command-line syntax to remember.
The footer lists the prefixes available in your current cluster context, so you see what you can search for.
Why a palette when you already have a sidebar?
Open the palette from anywhere with a keystroke: no menus to navigate, no long resource grids to scroll.
Type vm:web to list the matching VMs and containers of the selected cluster. You do not need to remember which node or storage hosts what.
Type >start and pick a VM: the VM starts. No need to open Resources, find the row and click its menu.
Lost a VM but know its IP? ip:192.168.1 lists every matching VM and container. The ID, the name and the Proxmox VE tags are searchable too.
| Prefix | Filters | Example | Scope |
|---|---|---|---|
| none | Modules, by name or description | backup |
All clusters |
vm: |
VM or container by ID, name or Proxmox VE tag | vm:100, vm:web |
Per cluster |
node: |
Node by name | node:pve1 |
Per cluster |
ip: |
VM or container by IPv4 address (also by ID, name or tag) | ip:192.168 |
Per cluster |
> |
Commands only | >start |
All clusters |
vm: and ip: return both QEMU virtual machines and LXC containers. For a VM, ip: reads the addresses reported by the guest agent; for a container, the addresses assigned now when it is running, otherwise the addresses in its configuration. The text after a prefix can be separated by a space: vm: debian works as vm:debian.
Commands open a dialog if they require parameters. PVE commands operate on the currently selected cluster.
| Command | Description | Parameters | Scope |
|---|---|---|---|
logout |
Log out the current user | - | All clusters |
start |
Start a stopped VM / Container | VM / CT (the picker lists the stopped guests) | Per cluster |
stop |
Stop a running VM / Container (stop, not shutdown) | VM / CT (the picker lists the running guests) | Per cluster |
restart |
Reboot a running VM / Container | VM / CT (the picker lists the running guests) | Per cluster |
console |
Open the NoVnc web console | VM / CT (the picker lists the running guests) | Per cluster |
create snapshot |
Create a VM / CT snapshot | VM / CT (the picker lists all the guests), Name (required), Description, Include RAM | Per cluster |
The command runs as soon as the dialog is confirmed, and its outcome is shown as a notification.
Each row shows an icon (color-coded by category), title, subtitle and status badges (running / stopped, hostname, OS, lock). The row ends with an indicator: open_in_new if the command opens a dialog, bolt for instant actions.
Result types: Module · VM · Node · Command.
The hostname, OS and IP address badges are filled in only by an ip: search. Selecting a module opens it. Selecting a VM or a node opens it in the Resources module.
No row is selected while you type: press ↓ to select one and Enter to run it, or click the row.

The query is split on spaces. A word starting with a known prefix (vm:, node:, ip:) selects what to search. Everything else, including the value after the prefix, is joined back into one search phrase.
> to list the commands.vm:, ip: or node:; commands need >.ip:, the IP addresses.vm:100 stopped doesn’t work. To filter by state, use the Resources module.node: and vm: falls back to node:.ip: reads the addresses of the running VMs through the QEMU Guest Agent, so it is slower than vm:. An address is found only on VMs that have the agent. Only IPv4 addresses are searched.
Console authentication: console requires the WEB API of the cluster to use Credential authentication. With API Token authentication the command only shows a notification asking to switch to the Credential access type.
Cluster context:
logout, the > command filter) are always there.The palette itself needs no permission. What it shows depends on the permissions of the user:
| Entry | Required permission |
|---|---|
| Module | The access permission of that module |
| VM, node | The cluster permissions of the user on that resource |
start, stop, restart |
VM Power Management on the selected cluster |
console |
VM Console on the selected cluster |
create snapshot |
VM Snapshot manager on the selected cluster |
logout |
none |
Proxmox VE privileges: the base (PVEAuditor) plus the following.
| Privilege | Used for |
|---|---|
VM.PowerMgmt | start, stop, restart |
VM.Console | console |
VM.Snapshot | create snapshot |
The base and the rules common to all modules are in Proxmox VE Permissions.