gbox
Cli and MCP for gbox. Enable AI agents to operate Android/Browser/Desktop like human. Go-based implementation.
Documentation
GBOX
GBOX provides environments for AI Agents to operate computer and mobile devices.
*Mobile Scenario:*
Your agents can use GBOX to develop/test android apps, or run apps on the Android to complete various tasks(mobile automation).
*Desktop Scenario:*
Your agents can use GBOX to operate desktop apps such as browser, terminal, VSCode, etc(desktop automation).
*MCP:*
You can also plug GBOX MCP to any Agent you like, such as Cursor, Claude Code. These agents will instantly get the ability to operate computer and mobile devices.
Installation
System Requirements
- MacOS
- Version: 10.15 or later
- Homebrew
> Note: Using gbox on other platforms, please check npm package @gbox.ai/cli for installation instructions. You can also login to GBOX.AI to use web-based dashboard.
Installation Steps
Quick Install (Recommended)
Option 1: Install GBOX CLI only
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bashThis installs just the GBOX CLI. Additional dependencies (like ADB, frpc, Appium) will be automatically installed when needed, or you can install them later using:
gbox setupOption 2: Install with all dependencies (one-step setup)
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash -s -- --with-depsThis installs everything at once:
- GBOX CLI
- Node.js and npm (if not present)
- Android Debug Bridge (ADB)
- FRP client (frpc)
- Appium server, drivers and plugins
Non-Interactive Mode:
# CLI only
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash -s -- -y
# With all dependencies
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash -s -- -y --with-depsAlternative Installation Methods
Via Homebrew (macOS):
brew install gboxVia npm (Linux/Windows):
npm install -g @gbox.ai/cliSetup Dependencies
If you installed GBOX CLI only, you can install all command dependencies anytime:
gbox setupOr install them interactively:
gbox setup -yUpdate GBOX CLI
Keep your GBOX CLI up to date:
Using the install script:
# Interactive update (will prompt if already installed)
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash
# Force update without prompt
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash -s -- --update
# Skip update even if newer version available
curl -fsSL https://raw.githubusercontent.com/babelcloud/gbox/main/install.sh | bash -s -- --update=falseUsing Homebrew (macOS):
brew upgrade gboxUsing npm (Linux/Windows):
npm update -g @gbox.ai/cliGet Started
# Login to gbox.ai
gbox login
# Connect your Android device (dependencies will be installed automatically if missing)
gbox device-connect
# Export MCP config and merge into Claude Code/Cursor
gbox mcp export --merge-to claude-code
gbox mcp export --merge-to cursorCommand Line Usage
Check GBOX CLI Reference for detailed usage.
SDK Usage
Check GBOX SDK Reference for detailed usage.
Use GBOX as a MCP Server(Login required)
Using GBOX CLI to configure MCP server to your Claude Code/Cursor:
# Export MCP config for Cursor
gbox mcp export --merge-to cursor
# Export MCP config for Claude Code
gbox mcp export --merge-to claude-code --scope projectOr copy paste the following content into your Claude Code/Cursor MCP config:
{
"mcpServers": {
"gbox-android": {
"command": "npx",
"args": [
"-y",
"@gbox.ai/mcp-server@latest"
]
}
}
}> Note:
> - Currently, GBOX MCP can only control Android environments.
> - If you need Cursor/Claude Code to control your local Android device, please check Register Local Device
Android MCP Use Cases
| Use Case | Demo |
|---|---|
| Claude Code Develop/Test Android App |  |
| Claude Code Compare Prices on eCommerce Apps |  |
Environments
Currently, GBOX supports the following environments:
- Android
- Linux Desktop/Browser
Android Environment
There are three types of Android environments, you can choose based on your needs:
1. Cloud Virtual Device:
Login to GBOX.AI to get a cloud virtual device. Best for testing and development.
2. Cloud Physical Device:
Login to GBOX.AI to get a cloud physical device. Cloud physical device is a real Android phone that you can use for production scenarios.
3. Local Physical Device:
Use your own physical device Register Local Device. Your local device can be any Android device that have Developer Mode enabled. Best for production scenarios and personal use.
Linux Desktop/Browser Environment
Login to GBOX.AI to get a Linux desktop/browser environment. Best for testing and development.
Develop gbox
Prerequisites
- Go 1.21 or later
- Make
- pnpm (via corepack)
- Node.js 16.13 or later
Build
# Build for current platform
make build
# Build for all platforms
make build-all
# Create distribution package
make distRunning Services
# MCP Server
cd packages/mcp-server && pnpm dev
# MCP Inspector
cd packages/mcp-server && pnpm inspectContributing
We welcome contributions! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.
1. Fork the repository
2. Create your feature branch (`git checkout -b username/feature-name`)
3. Commit your changes (`git commit -m 'Add some feature'`)
4. Push to the branch (`git push origin username/feature-name`)
5. Open a Pull Request
Things to Know about Dev and Debug Locally
License
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
Frequently asked questions
What is gbox?
gbox is Cli and MCP for gbox. Enable AI agents to operate Android/Browser/Desktop like human. Go-based implementation.
How do I install gbox?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is gbox open source?
Yes — it is hosted on GitHub at https://github.com/babelcloud/gbox and has 145 stars.
Related MCP tools
A fully functional MCP server and CLI for YouTube Go-based implementation.
:robot: The free, Open Source alternative to OpenAI, Claude and others. Self-hosted and local-first. Drop-in replacement for OpenAI, running on consumer-gra...
MCP Toolbox for Databases is an open source MCP server for databases. Go-based implementation. Trusted by 10900+ developers.
WhatsApp MCP server Go-based implementation. Trusted by 4900+ developers. Trusted by 4900+ developers. Trusted by 4900+ developers.
Query anything (GitHub, Notion, +40 more) with SQL and let LLMs (ChatGPT, Claude) connect to using MCP Go-based implementation. Trusted by 1300+ developers.
ToolHive makes deploying MCP servers easy, secure and fun Go-based implementation. Trusted by 1300+ developers. Trusted by 1300+ developers.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP