rustdesk-mcp-bridge

MCP server that exposes local desktop automation tools: screen capture, mouse movement/clicks, and keyboard input. Designed to be driven by an AI agent so it can operate remote desktops (for example via RustDesk) by running this server on the target machine.

Requirements

• Python 3.11+

• A graphical session on the target machine (pyautogui needs an interactive desktop on Windows; on Linux it needs an X11/Wayland session).

Related MCP server: vision-input

Installation

python3 -m venv .venv
.venv/bin/python -m pip install -e '.[dev]'

Usage

Run the MCP server over stdio:

rustdesk-mcp-bridge

Available MCP tools:

• get_screen_size — current screen resolution.

• capture_screen — base64 PNG/JPEG screenshot of the full screen or a region.

• move_mouse — move the cursor to absolute coordinates.

• mouse_click — click at absolute coordinates or at the current position.

• mouse_down / mouse_up — press/release a mouse button (drag support).

• type_text — type a Unicode string.

• send_key — press a single key or a combination (e.g. ctrl+a).

Windows / RustDesk target setup

1. Install the package on the Windows target (e.g. nzxt).

2. Run rustdesk-mcp-bridge from an interactive PowerShell session on the target desktop (not a plain SSH session), so pyautogui can send input.

3. Connect your MCP client to the server's stdio. If you run it over SSH, use a terminal multiplexer launched from the interactive desktop.

Development

.venv/bin/python -m pytest
.venv/bin/ruff check .
.venv/bin/mypy src

Security

Never commit RustDesk IDs, passwords, or credentials. This server can execute arbitrary mouse and keyboard actions: only run it in trusted environments and protect its MCP transport.

License

MIT