Opencu

[OpenCU]

OpenCU

Computer use for any AI agent, on your own machine.

https://github.com/user-attachments/assets/5da04b21-47fe-44c4-9d3d-e2b3a2e9a9c6

_{Re-created in code from a real session.}

OpenCU is an MCP server that lets Claude Code, Codex or any other MCP client read
and operate the apps on macOS, Windows, Linux and Android. It runs locally, needs no account
or model key, and the person at the keyboard can stop it at any time.

[Latest release] [macOS, Windows, Linux, Android] [MIT license]

Install · Connect · Measured · Limitations · 简体中文

Why OpenCU

It works beside you. On macOS, input goes to the target app's process and window. Your cursor stays where it is and the app in front keeps the keyboard.

It can have a screen of its own. Start the agent desktop and the agent works on a virtual display (macOS) or in a separate session (Windows), which you watch in an ordinary window.

You stay in charge. You can always stop it: the Stop button on macOS, a physical Escape on Windows and Linux, three presses of volume-down on Android. On Windows, Linux and Android it will not take the foreground within 10 seconds of your last input. A stopped connection never restarts on its own.

It is measured. On Windows tasks that can only be done through the UI, the same agent passed 7 of 8 trials with OpenCU and 2 of 8 with a shell and screenshots. Details below.

It is yours. MIT licensed, with no account, no proprietary runtime and nothing hosted. The agent you already use brings the model.

Install

Windows · PowerShell, as your normal user, no administrator rights

irm https://github.com/Daily-AC/opencu/releases/latest/download/install.ps1 | iex

Linux · x86_64 with glibc, X11 desktops

curl -fsSL https://github.com/Daily-AC/opencu/releases/latest/download/install.sh | sh

macOS · Apple silicon, macOS 14 or later, Node.js 22+

npm install -g @z10/opencuopencu setup    # grant Accessibility and Screen Recording

Android · preview, Android 14 or later, driven from a computer over adb (USB or wireless debugging)

npm install -g @z10/opencu    # Apple silicon Macs, with adb; other computers build from source, see the Android guideopencu android doctor

To check an installation, run opencu doctor (on Windows and Linux, the command from the table below). Platform guides: macOS · Windows · Linux · Android. Connecting your own host to Android: Android integration.

Connect your agent

OpenCU is a stdio MCP server. The command to register is:

# Claude Codeclaude mcp add --scope user opencu -- opencu
# Codexcodex mcp add opencu -- opencu

On Windows, Linux and Android, use the command from the table instead of opencu.

Any other MCP client (on Android, add "args": ["android"]):

{  "mcpServers": {    "opencu": { "command": "opencu" }  }}

Then ask your agent to do something in an app. It starts by reading the app with get_app_state.

What the agent can do

select_text is macOS and Linux only. Linux has no agent_desktop.

Input requires a fresh observation of the same app, and every error starts with a stable code such as STOPPED_BY_PERSON. Details are in Tools and the engine contract.

Measured

With and without OpenCU. The same agent and model, with a shell and screenshots alone or with OpenCU added:

On the scriptable tasks the shell-only agent wrote LibreOffice and OBS scripts instead of using the UI; two of OpenCU's three failures there were a modal-dialog defect fixed in 0.3.2. Evidence: Windows, Linux.

Against open-computer-use. The same agents, models and tasks with OpenCU and with iFurySt/open-computer-use 1.1.2 at its default settings:

• Linux: the two Firefox tasks took OpenCU 244 and 96 seconds (median) and open-computer-use 872 and 552. open-computer-use did not list Firefox until the agent turned on the AT-SPI bus itself, and in one trial its set_value and type_text left the Firefox entries empty without an error.
• Windows: in all eight open-computer-use trials the agent also wrote its own Win32 input code, and its four passes completed input that way. It wrote none with OpenCU. open-computer-use ran behind a bridge on the VM (1.5 to 7 seconds per call), so its Windows timings are not comparable.
• macOS: both passed the same four tasks. On those, the agent took 544 seconds in total with OpenCU and 1,498 with open-computer-use; median click time was 292 ms against 2,029 ms. The OpenCU build in that VM was 0.2.0.

The samples are small and each platform used one model. We did not test open-computer-use's locked-screen mode, its background input on covered Chrome windows, or its global pointer setting (evidence).

Platforms

Stopping it

• macOS: the V in the menu bar has Stop. In agent desktop mode, the preview's Stop button does the same.
• Windows and Linux: a physical Escape stops input for the rest of the connection, and your own input during an action stops that action. The Windows agent desktop also has a Stop button.
• Android: press volume-down three times within 1.5 seconds. While the agent works on its background screen, the optional phone app's status island also has Stop. Touching the main screen during an action there stops that action.
• A person's stop is final for the connection. Agents must never reconnect around it.

OpenCU is a person-facing brake, not a sandbox: the agent's apps share your account and files, and any local process that can run opencu uses its permissions.

Limitations

• macOS: Apple silicon only. The package is signed with OpenCU's own certificate and is not notarized, so install it with npm, not a browser download. macOS asks about once a month whether OpenCU may keep capturing the screen. It cannot operate a locked Mac.
• Windows: the package is unsigned, so SmartScreen or Defender may warn. Outside the agent desktop, input uses the foreground.
• Linux: X11 only; Wayland is refused. Input uses the foreground.
• Android: a preview, tested on one phone (OPPO PGBM10, Android 14). It runs from a computer over adb; the npm package installs only on Apple silicon Macs, and other computers build from source. The phone app is optional and only shows status, a preview and Stop.
• Custom-drawn apps such as Blender, and windows in the Stage Manager strip, need the foreground.

The full list, with measurements, is in Limitations.

Documentation

• Platform guides: macOS, Windows, Linux, Android
• Tools and the engine contract, with the Windows and Linux contracts
• Host integration, for programs that start OpenCU on an agent's behalf
• Limitations and measured behavior

Development

npm testpython3 scripts/test-native.pynode scripts/verify-gui.mjs

verify-gui.mjs builds two disposable AppKit fixtures, keeps one in the foreground and operates the other in the background through the MCP server. On Windows, node scripts/verify-gui-win.mjs uses disposable WinForms and Chrome fixtures and takes the foreground.

python3 scripts/package-opencu.py builds a self-contained archive with Node and the MCP dependencies for your own Macs, signed with your local identity, and python3 scripts/verify-release.py checks it offline.

License

Original OpenCU code is available under the MIT License. Bundled open source dependencies retain their own licenses. No license for proprietary third party software is included or implied.