deskbrid

deskbrid

deskbrid.patchhive.dev

mcp-name: io.github.coe0718/deskbrid

📖 Documentation | API Reference | Architecture

The HAL your Linux desktop agents are missing.

Deskbrid is a single Rust binary that auto-detects your desktop environment and wraps it into a JSON-over-Unix-socket protocol. GNOME, Hyprland, KDE, COSMIC, Sway, Niri, Wayfire, Labwc, Cinnamon, MATE — one daemon, one protocol, one binary.

# Human
deskbrid windows list
deskbrid clipboard read

# Agent (same socket)
{"action": "windows.list"}  →  [{"title": "VS Code", "app_id": "code", ...}]

Table of Contents

• Why Deskbrid
• Supported Desktops
• Installation
• Quick Start
• Features
• Protocol
• Python Client
• MCP Integration

Why Deskbrid

Every major AI lab is racing to ship desktop agents. AppleScript gives macOS agents native control. Windows has UI Automation. Linux has xdotool — which breaks on Wayland, the default display protocol for every major distro.

Deskbrid fills that gap. It auto-detects your compositor and loads the right backend — GNOME (Mutter RemoteDesktop DBus), Hyprland (hyprctl + ydotool + grim), KDE (KWin D-Bus + ydotool + spectacle), wlroots-style compositors, or shared X11. Same binary, same protocol, same socket.

Dashboard

Deskbrid ships with a built-in web dashboard at localhost:20129 — system info, monitors, windows, network, audio, clipboard, and an audit log of agent actions, all live:

🔴 Live Demo →

Supported Desktops

Deskbrid auto-detects your desktop at startup ($XDG_CURRENT_DESKTOP → process scan → GNOME fallback). No config files, no flags.

See DE Test Matrix for per-action compatibility across all desktops — every action, every compositor, tested on real hardware.

Installation

One-liner install (recommended):

bash <(curl -fsSL https://deskbrid.patchhive.dev/install.sh)

Auto-detects your distro and desktop environment, installs dependencies, sets up uinput, and downloads the binary.

Manual installation:

Download the latest release binary from the releases page:

curl -LO https://github.com/coe0718/deskbrid/releases/latest/download/deskbrid
chmod +x deskbrid
sudo mv deskbrid /usr/local/bin/

Or build from source:

git clone https://github.com/coe0718/deskbrid
cd deskbrid
cargo build --release
sudo cp target/release/deskbrid /usr/local/bin/

Desktop Setup

GNOME:

sudo apt install -y grim wl-clipboard python3-gi gstreamer1.0-tools gstreamer1.0-pipewire
deskbrid setup

Hyprland (and other standalone Wayland compositors — Sway, Niri, Wayfire, Labwc):

sudo pacman -S grim wl-clipboard ydotool
echo 'KERNEL=="uinput", GROUP="input", MODE="0660"' | sudo tee /etc/udev/rules.d/99-input.rules
sudo usermod -aG input $USER

⚠️ Standalone Wayland compositors don't ship a notification daemon. Deskbrid's notify send will hang without one. Install dunst, mako, or swaync and add it to your compositor's autostart.

KDE Plasma:

sudo apt install spectacle imagemagick wl-clipboard ydotool

Quick Start

deskbrid daemon &

deskbrid windows list          # List open windows
deskbrid clipboard read        # Read clipboard
deskbrid screenshot            # Take screenshot
deskbrid system info           # Get system info
deskbrid windows focus --app code  # Focus VS Code
deskbrid input keyboard type "Hello!"  # Type text

Features

Windows & Workspaces

Input & Clipboard

Screenshots & Media

System & Services

Network & Bluetooth

Protocol

Deskbrid uses JSON-over-Unix-socket. See PROTOCOL.md for the complete specification.

→ {"action": "windows.list"}
← {"type": "response", "status": "ok", "data": [{"title": "VS Code", ...}]}

→ {"action": "windows.focus", "window_id": "code"}
← {"type": "response", "status": "ok"}

Events

Subscribe to real-time updates:

{"action": "subscribe", "events": ["file.*"]}

Python Client

from deskbrid import Deskbrid

client = Deskbrid()

# List and focus VS Code
windows = client.windows_list()
code_window = next((w for w in windows if w.app_id == 'code'), None)
if code_window:
    client.focus_window(app_id='code')
    client.type_text("Fixed the bug!\n")

# Subscribe to events
@client.on("file.*")
def on_file_change(event):
    print(f"File changed: {event['path']}")

MCP Integration

Deskbrid exposes a full Model Context Protocol server for AI coding tools:

deskbrid mcp

Claude Desktop (~/.config/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "deskbrid": {
      "command": "/usr/local/bin/deskbrid",
      "args": ["mcp"]
    }
  }
}

Available MCP tools (20+):

• list_windows, focus_window
• type_text, press_keys, mouse_move, mouse_click
• screenshot, clipboard_read, clipboard_write
• list_apps, get_accessibility_tree
• perform_action, set_element_value, get_element_text, click_element
• doctor, setup_accessibility, capabilities

Compared to Alternatives

License

MIT

How This Started

Deskbrid began with Tuck — an autonomous agent that needed to control a real Linux desktop. When the community asked for Hyprland support, Tuck asked Jeremy for a bare Arch Linux box with SSH and sudo. He installed Hyprland himself and built the backend from inside the environment he just configured.

The first working demo was a Telegram message: Tuck focused a window and typed "Hello from the other side" in under 60 seconds. That moment — an agent controlling a real desktop through a Unix socket — became Deskbrid. It's built for agents first: same protocol for humans on the CLI, same socket for AIs, one binary that works everywhere.