Agent-OS

Agent-OS is a comprehensive local agentic system designed to bridge the gap between secure, local Linux administration and modern AI capabilities. It consolidates multiple tools—an advanced terminal interface, a safety-focused "Turtle Shell" execution layer, and a web-based chat frontend—into a single, cohesive operating environment for your AI interactions.

🚀 Features

• Local-First Architecture: Powered by local LLMs (via Ollama) for privacy and offline capability.
• Turtle Shell Safety Layer: A robust permission and safety verification system that acts as a guardrail between the AI and your system kernel.
• Hybrid Intelligence: Seamlessly route complex queries to cloud models (like Google Gemini) while keeping sensitive system operations local.
• Multi-Interface Access:
  • CLI REPL: A rich terminal user interface for direct, low-latency interaction.
  • Web Frontend: A modern, responsive React/Vite interface for a polished chat experience.
• Context-Aware: Intelligently manages context across sessions to maintain continuity without overwhelming the model.

🛠️ Prerequisites

Before running Agent-OS, ensure you have the following installed:

• Python 3.10+
• Node.js & npm (for the frontend)
• Ollama: Must be installed and running (ollama serve).
  • Note: Ensure you have pulled the necessary models (e.g., llama3, mistral).

📥 Installation

1. Clone the repository:

   git clone https://github.com/your-username/agent-os.git
   cd agent-os

2. Set up the Python Environment:

   python3 -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt

3. Set up the Frontend:

   cd frontend
   npm install
   cd ..

4. Configure Environment: Create a .env file in the root directory with your API keys (if using cloud fallbacks):

   GOOGLE_API_KEY=your_key_here

🚦 Quick Start

1. Start Ollama

Ensure your local LLM backend is running. Open a terminal and run:

ollama serve

2. Launch Agent-OS (Backend + REPL)

In a new terminal:

source venv/bin/activate
python main.py

This starts the core system and the text-based interface.

3. Launch the Frontend (Optional)

To use the web interface:

cd frontend
npm run dev

Open your browser to http://localhost:5173.

🛡️ Safety & Security

Agent-OS implements a Zero-Trust approach to system commands. The Turtle Shell module analyzes every generated command for potential destruction (file deletion, system config changes) and requires explicit user confirmation for high-risk actions.

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for details on how to submit pull requests, report issues, and suggest improvements.

📄 License

Distributed under the MIT License. See LICENSE for more information.

---

_{Developed with the assistance of Google Gemini 3 Pro and Claude 4.5 Opus.}