open-zcode-computer-use
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPEN_ZCODE_CUA_NATIVE | Yes | Absolute path to the native OpenZCodeNative binary required by the MCP server. | |
| OPEN_ZCODE_CUA_ALLOW_GLOBAL_POINTER | No | Set to '1' to opt into physical-pointer fallback for raw move and split mouse-down/up. Defaults to disabled when unset. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| list_appsA | List currently running applications. |
| open_applicationA | Resolve or launch an application without taking foreground focus unless activate is true. |
| list_windowsB | List windows belonging to an application. |
| get_app_stateB | Capture an accessibility-first application state and optionally its window image. |
| screenshotA | Capture the selected display. |
| zoomB | Crop a region from an existing frame or around an element target. |
| list_displaysA | List attached displays and their coordinate spaces. |
| switch_displayA | Select the display used by screenshot. |
| cursor_positionA | Read the physical cursor position. |
| left_clickB | Click a bound element or frame coordinate. |
| double_clickB | Double-click a bound element or frame coordinate. |
| triple_clickC | Triple-click a bound element or frame coordinate. |
| right_clickB | Right-click a bound element or frame coordinate. |
| middle_clickC | Middle-click a bound element or frame coordinate. |
| scrollC | Scroll at a bound element or frame coordinate. |
| left_click_dragC | Drag from one bound target to another. |
| mouse_moveA | Move a synthesized pointer without moving the user's physical cursor. |
| left_mouse_downC | Begin a scoped left-button hold. |
| left_mouse_upA | Release the scoped left-button hold. |
| typeB | Type text into a bound element or the focused element of a scoped app. |
| set_valueC | Set an accessibility element value. |
| select_textC | Select all or a UTF-16 range in a text element. |
| keyA | Send a key or chord to a scoped application. |
| hold_keyB | Hold a key or chord for up to 30 seconds. |
| perform_actionC | Run one of the accessibility actions advertised by an element. |
| request_accessB | Read permission and runtime readiness without prompting. |
| stop_computer_controlB | Latch the session kill switch and release held input. |
| waitA | Pause for up to 30 seconds. |
| read_clipboardA | Read plain text from the system clipboard. |
| write_clipboardA | Replace the system clipboard with plain text. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 30 tools
Every tool targets a distinct action or resource category: display/window/app listing, mouse variants, keyboard input, accessibility operations, clipboard access, and control flow. The closest pairs like key vs hold_key and type vs set_value are clearly differentiated by duration and input mechanism.
Most names follow a predictable lower_snake_case verb_noun pattern such as list_windows, open_application, set_value, and read_clipboard. A few outliers like key, screenshot, and cursor_position break the pattern slightly but remain recognizable and consistent in style.
With 30 tools, the server feels over-scoped for a single computer-use entry point. Many click variants and keyboard/pointer hold primitives could be parameterized or consolidated without losing capability, though the broad domain justifies some of the count.
The surface covers the main computer-use workflows: display selection, screenshots, app/window introspection, mouse and keyboard input, accessibility actions, and clipboard management. Notable minor gaps include no app termination or active window management, but agents can work around these via existing accessibility actions.