open-computer-use
github.com
- Free plan
- Windows
- macOS
- Linux
- API
- Self-hosted
What open-computer-use is
Open Computer Use is a free, local MCP server that allows AI agents to inspect and operate desktop applications through accessibility interfaces. Its nine core tools cover app discovery, accessibility-tree and screenshot capture, clicking, scrolling, dragging, typing, key presses, secondary actions and setting control values. The project provides setup instructions for Codex CLI and App, Claude Code, Gemini CLI and opencode, as well as manual configuration for other MCP clients. It also includes a ZCode plugin for installing the skill and connecting an MCP server. The project says it runs on the user's machine and uses accessibility interfaces for core app actions without taking over the real mouse and keyboard. Password managers are refused; app launching, focus changes and global input injection require explicit environment-variable gates. macOS 14 or later needs Accessibility and Screen Recording permissions. Windows and Linux require a signed-in desktop session, and Linux also requires AT-SPI2. The software is MIT-licensed and supports macOS, Windows and Linux.
Who it suits
It suits people building AI-agent workflows that need to interact with desktop applications on an authorized local machine. Users should be comfortable configuring an MCP client and granting required system permissions.
Good to know, and what to watch
- Runs locally on macOS, Windows and Linux
- Nine tools cover screen inspection and desktop actions
- Setup guidance covers several MCP clients
- MIT-licensed and free
- macOS 14 or later requires two permissions
- Linux requires AT-SPI2
- Window-level tools are currently available only on Windows
Plans
Every plan and priceOpen source
Open source
The verdict
Open Computer Use provides a local route for AI agents to work through desktop accessibility interfaces, with setup guidance for several clients. It handles high-privilege automation in a signed-in session, so use is intended for authorized machines.
What the maker publishes
- Desktop actions
- Its nine core tools support app discovery, accessibility-tree and screenshot capture, clicking, scrolling, dragging, typing, key presses, secondary actions, and setting control values. source
- Integrations
- The project documents setup commands for Codex CLI and App, Claude Code, Gemini CLI, and opencode, and supports manual configuration for other MCP clients. source
- License
- The repository is distributed under the MIT license. source
- Local operation
- The project says it runs entirely on the user's machine and uses the accessibility layer without taking over the real mouse and keyboard for its core app actions. source
- Permissions
- On macOS 14 or later, the project requires Accessibility and Screen Recording permissions; Windows and Linux work in a signed-in desktop session, with Linux requiring AT-SPI2. source
- Platform limitation
- The additional window-level tools are currently available on Windows, with macOS and Linux support in progress. source
- Platforms
- The runtime supports macOS, Windows, and Linux, with nine core tools available across all three. source
- Purpose
- Open Computer Use is a local MCP server that lets AI agents inspect and operate desktop apps through accessibility interfaces. source
- Safety controls
- Password managers are refused, while app launching, focus changes, and global input injection require explicit environment-variable gates. source
- Security policy
- The security policy says fixes are provided for the latest npm release and asks users to report suspected vulnerabilities through GitHub's private vulnerability reporting channel. source
- Security scope
- The policy notes that the tool performs high-privilege accessibility and input automation in the local signed-in desktop session and should only be used on authorized machines. source
- Support and audience
- The project provides installation and troubleshooting references and invites contributions; its security policy says fixes are delivered through the regular release process. source
- ZCode plugin
- The repository includes a ZCode plugin that installs the skill and connects an MCP server. source
Checked against the maker's pages on 9 Oct 2026.