codex-computer-use-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CODEX_COMPUTER_USE_HOME | No | Override the default state directory (~/.codex-computer-use-mcp) with a custom path. |
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 | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| list_appsA | List the apps on this computer. Returns the set of apps that are currently running, as well as any that have been used in the last 14 days, including details on usage frequency |
| get_app_stateA | Start an app use session if needed, then get the state of the app's key window and return a screenshot and accessibility tree. This must be called once per assistant turn before interacting with the app |
| clickB | Click an element by index or pixel coordinates from screenshot |
| perform_secondary_actionC | Invoke a secondary accessibility action exposed by an element |
| set_valueB | Set the value of a settable accessibility element |
| select_textA | Select text inside a text element, or place the text cursor before or after it. Provide text exactly as it appears in the accessibility tree, including any Markdown formatting. If the text is not unique, provide surrounding prefix or suffix text to disambiguate it. |
| scrollB | Scroll an element in a direction by a number of pages |
| dragB | Drag from one point to another using pixel coordinates |
| press_keyB | Press a key or key-combination on the keyboard, including modifier and navigation keys. This supports xdotool's |
| type_textA | Type literal text using keyboard input |
| computer_use_statusC | Show Computer Use status. |
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 11 tools
Each tool maps to a distinct UI interaction such as clicking, typing, scrolling, dragging, or value setting, so an agent can generally pick the right one. The main overlap risk is between `type_text` and `press_key`, since both are keyboard interactions, and `perform_secondary_action` versus `click` could be confused when a secondary action is not clearly described.
Most tools follow a lowercase snake_case verb-first pattern such as `list_apps`, `get_app_state`, `type_text`, and `select_text`. The consistency is marred slightly by `computer_use_status`, which reads as a noun phrase rather than an imperative verb like `get_computer_use_status`, and by a few bare verbs like `click` and `drag`.
With 11 tools, the set is well-scoped for the stated purpose of computer use through screenshots, accessibility trees, keyboard input, and pointer actions. Every tool covers a needed capability without bloating the surface.
The tools collectively cover common end-to-end GUI workflows: observe state, click, type, press shortcuts, scroll, drag, set values, and select text. There are minor gaps such as no dedicated hover, app-launch, or app-close tool, but those can often be worked around with the existing pointer and keyboard actions.