computer-use-linux

Overview of computer-use-linux

What it is

computer-use-linux is a Rust MCP server and CLI for native Linux desktop control. It gives AI agents structured control of a local Linux desktop through semantic AT-SPI accessibility trees, bounded screenshots, compositor-aware window targeting, Wayland/X11 input synthesis, single-round-trip post-action verification, and a click-through on-screen activity indicator. It supports GNOME, KDE Plasma/KWin, Hyprland, niri, COSMIC, i3, and X11/EWMH.

Key features

• Semantic AT-SPI accessibility trees for element inspection and interaction - Compositor-aware window targeting across multiple desktop environments - Wayland input synthesis via /dev/uinput or org.freedesktop.portal.RemoteDesktop, with fallbacks - Single-round-trip post-action verification (screenshot + updated accessibility tree after each action)
• Click-through on-screen activity indicator that hides automatically before screenshots - One JSON readiness report via the doctor tool

Use cases

AI agents controlling a local Linux desktop for tasks such as clicking, typing, dragging, taking screenshots, and inspecting accessibility trees. It is also used for desktop automation and testing.

Getting started

Install the CLI globally via npm or from source, then verify desktop readiness: bash npm install -g @agent-sh/computer-use-linux # or from source: git clone https://github.com/agent-sh/computer-use-linux && cd computer-use-linux && ./install.sh computer-use-linux doctor | jq .readiness computer-use-linux setup computer-use-linux setup-window-targeting For agent-specific setups (Claude Code, Codex CLI, Pi Coding Agent, Hermes Agent), see the repository README.

License and pricing

MIT license, open source.

Based on the project's public README and repository metadata.