Buddy-OS is a Linux-based operating system designed to provide a fully autonomous AI developer assistant that can perform any task a human developer could do on the system.
- Full System Access: Buddy AI can launch applications, write files, run scripts, manage containers, and control the graphical interface.
- Buddy Copilot: GUI chat interface at bottom-right of desktop for text-based interaction.
- Hey Buddy Voice Control: Voice activation service that listens for "Hey Buddy" wake word.
- Policy System: Configurable access policies with "restricted" preset for developer mode.
- GUI Automation: Mouse, keyboard, screen capture, and window management capabilities.
- Docker Control: Manage containers and images.
- Network Administration: Configure network interfaces and monitor network status.
- Shell Execution: Run system commands with full privileges.
- Linux system (Ubuntu/Debian recommended)
- Python 3.8+
- Git
- Clone the repository:
git clone https://github.com/yourusername/Buddy-OS.git
cd Buddy-OS- Install dependencies:
./build_all.sh- Start Buddy AI:
./start_buddy.sh- Test Buddy AI capabilities:
./scripts/dev/test_all_buddy.shOnce Buddy AI is running:
- Use the Buddy Copilot GUI at bottom-right of desktop to interact with Buddy via text
- Say "Hey Buddy" to activate voice control (voice service running in background)
- Buddy will execute tasks and provide feedback in real-time
broker/: Core action execution daemon with HTTP APIbuddy-copilot/: GUI chat interface for user interactionbuddy-voice/: Voice activation service for hands-free controlscripts/: Build and development scriptsnotes/: Development decisions and documentation
Run the test script to verify all capabilities:
./scripts/dev/test_all_buddy.shBuddy-OS implements a policy system to control access:
- Default policy allows "restricted" access in developer mode
- Blacklists sensitive directories (
.ssh,.gnupg) - Capabilities can be disabled via environment variables
- Action execution is logged for audit purposes
- Consent and re-auth requirements enforced for high-risk actions
Contributions are welcome! Please submit issues or pull requests with clear descriptions of the problem or feature.
This project is licensed under the MIT License - see the LICENSE file for details.