Desktop context menu

The desktop menu opens at the pointer when you right-click an empty part of the desktop. Unlike the menu bar, it is yours to write: a short text file with a label and an action on each line, so that what you open most is one right-click away.

What a menu of your own is for

For you:

• Shortcuts to your own things. One line opens a project's folder in the Workspace, a file in the text editor, a VM on its Terminal tab, an app or a web address.
• A command-line tool in a window. terminal btop opens btop in a window of its own, and any other command on the host works the same way.
• Lists that fill themselves. @vms, @apps and @windows show this desktop's VMs, apps and open windows each time the menu opens. Each VM brings a submenu to open, start, stop or restart it.
• Your order and your words. Rename and reorder the items, group them in submenus, and leave out what you never use.
• The same menu everywhere. A saved menu reaches the other open desktops and syncs to joined nodes.

For a coding agent, such as Claude Code or Codex:

• It can change the menu for you. An agent reads the menu with GET /v1/ui/menu and replaces it with PUT /v1/ui/menu. The daemon checks what it sends as it checks Save, so a file with a mistake never replaces your menu.
• Only when you ask. The Agent Skill Guide tells an agent to change your menu only when you ask it to. "Add the demo VM's terminal to my desktop menu" is such a request.

Open the menu

Right-click an empty part of the desktop, or long-press it on a touch screen. A quick two-finger tap also opens the menu; on a phone this works inside an ordinary window too, away from controls, menus and dialogs. It is a NeXT-style menu: it opens where the pointer is, with its submenus beside it.

The factory menu offers New VM, Terminal, Workspace, lists of VMs, apps and windows, Show All Windows, Show Desktop, tools, Cloudflare and Help. Its last item is Customize…, which opens the menu editor.

Edit and save

1. Open the menu and choose Customize…. The Desktop Menu window opens on the menu's text.
2. Edit the text. In this window the Tab key types a tab.
3. Click Save, or press Command-S on a Mac or Ctrl-S elsewhere.

Save checks the format first: an error identifies and selects the bad line, and your current menu stays in place until the file is valid.

Until you save a menu of your own, the editor shows the factory menu. Its opening comments describe the format and name every action.

The scrollbars run along the document's right and bottom edges; drag the resize tile where they meet to make the editor wider or taller.

Where the menu is kept

Your menu is stored as System :menu.txt (normally ~/.exe/appdata/System/menu.txt). The file is there only once you have saved a menu of your own.

A saved change reaches other open desktops and syncs to joined nodes; see Joining desks together. Each desktop fills the automatic lists from its own VMs, installed apps and open windows.

A menu of your own keeps its contents when the factory menu changes, so a new shortcut in the factory menu does not appear in yours by itself. Add it to your own file with Customize….

Let an agent change the menu

An agent, or a script of your own, reaches the same menu over HTTP:

• GET /v1/ui/menu answers with the menu's text in text, the parsed menu in menu, and default, which is true while the factory menu is in use.
• PUT /v1/ui/menu takes the new text as the request body, in plain text, and applies it as Save does. A file with a mistake changes nothing and is answered 400 with {"error": "line N: …"}. An empty body restores the factory menu.

$ curl -s http://127.0.0.1:7777/v1/ui/menu
$ curl -s -X PUT --data-binary @menu.txt http://127.0.0.1:7777/v1/ui/menu

When api_token is set, both calls need the Authorization: Bearer <token> header; see Configuration.

Menu format

Write one item per line. Separate the label from its action with a tab or at least two spaces. Single spaces are fine inside a label; two spaces end it. Action names, list names and VM tab names are lowercase and case-sensitive.

In the format examples below, <TAB> means a tab character, not the literal text <TAB>. The complete example menu further down uses spaces and can be copied directly.

A line that starts with # is a comment and one that starts with @ is read as a list, so a label cannot begin with either. Put comments on their own lines, rather than after an action.

Indent submenu children consistently, for example with two spaces per level. The top level has no indentation. When returning to a parent level, use exactly its earlier indentation. Only a plain submenu heading can have indented children; an action or automatic list cannot.

All supported actions

The following table lists every supported action. In the syntax column, angle brackets mark a required value and square brackets an optional one; replace these placeholders rather than typing the brackets. Each action needs a label before it, as in My terminal<TAB>terminal.

There is no separate macos9, macosx or macosxintel action in the customization format; use app macos9, app macosx and app macosxintel.

Arguments, paths and app names

Except for terminal, arguments are separated by whitespace: quotation marks do not combine words into one argument. A Workspace path or app name containing spaces therefore cannot be used directly in an edit, workspace or app item. Use workspace to browse to that file or folder, or @apps to list apps by their display names.

For workspace and edit, use relative paths such as Projects/demo or Projects/demo/README.md. Absolute paths and paths that escape the Workspace are rejected.

For built-in apps, use their lowercase IDs: app macos9, app macosx, app hub, app bluepencil, app dict or app analytics. Other apps use their installed folder name; see Apps.

Run a command in a window

The command after terminal is kept as one command line, including spaces, quotes and shell operators. For example, Monitor<TAB>terminal btop opens btop on the host if it is installed. To reach a VM's shell instead, use vm <name> term.

On Linux and macOS the command runs in the host user's login shell; on Windows it runs in PowerShell. The terminal session ends when the command exits.

A command's window opens at 80×31 characters at least — what btop needs with its GPU box, and room enough for any other tool — and cannot be dragged smaller. It remembers its size and position, so the next open of the same command puts it back where it was left.

Choosing the item again while the command runs brings its window forward; after the command has exited, it runs it again in the same window's place.

VM tabs

Use these values as the optional second argument to vm. For example, Demo terminal<TAB>vm demo term opens the Terminal tab of the VM named demo.

Automatic lists

These lists fill themselves when you open the menu. Use a list name alone to insert its entries in place, or give it a label to make a submenu. They take no arguments. An empty list shows a disabled No VMs, No Apps or No Windows entry.

Each @vms submenu offers Open, Open Terminal, btop, Chat with this VM, Start or Stop, Restart, Copy IP, Expose Port… and Publish to GitHub…. Availability follows the VM's state, its IP and Chat availability. Delete is not included.

These operations are supplied by @vms; names such as start, stop and restart are not standalone customization actions.

@apps leaves out Welcome, which has no desktop icon either; the welcome action opens it.

Example menu

This example uses two spaces between a label and its action, and two spaces of indentation for each submenu level. Replace demo and the Workspace paths with your own VM and files; install btop on the host to use the monitor shortcut.

# My desktop menu
New VM…  newvm
Terminal  terminal
Workspace  workspace
-
Projects
  Demo folder  workspace Projects/demo
  Demo README  edit Projects/demo/README.md
  Demo terminal  vm demo term
  Demo notes  vm demo notes
Tools
  Monitor  terminal btop
  Claude Code  claude
  Codex  codex
  Mac OS 9  app macos9
  Hub  app hub
Virtual Machines  @vms
Applications  @apps
Windows  @windows
Show All Windows  showall
Show Desktop  showdesk
-
Help
  Using exe  docs
  Online documentation  url https://exe.v2core.com/docs/
Customize…  customize

Limits

The menu may contain up to 300 entries, counting headings, actions, automatic-list entries in the file and separators across all levels. Expanded list contents do not count toward that limit.

Labels may be up to 60 characters; the file must be UTF-8 and at most 64 KiB. You can nest three submenu levels below the root (four levels in total).

Recover or reset the menu

If you remove customize, the desktop appends Customize… so you can always get back to the editor. If a stored or synced file cannot be parsed, the desktop uses the factory menu and adds The menu file has an error… to open the editor on the problem.

To reset, choose Restore Defaults and confirm, or clear the editor completely and Save. This removes the custom menu from every desktop sharing the desk. Only an empty or whitespace-only file restores defaults; a file containing just comments is still a custom menu.