AgentOS

AgentOS is a powerful platform for building, executing, and managing AI agents. It provides a comprehensive environment with a SOP (Standard Operating Procedure) engine, sandboxed code execution, browser automation, and a marketplace for sharing agent capabilities.

🚀 Features

• Agent Workbench: A visual environment to design, test, and refine your AI agents.
• SOP Engine: Define complex workflows using Standard Operating Procedures. Supports various node types including:
  • Interaction: Chat with users or other agents.
  • Reasoning: LLM-based decision making.
  • Action: Execute specific tasks or tools.
  • Human: Request human input or approval.
  • Sub-SOP: Nest workflows for modularity.
• Sandboxed Execution: Securely run Python code generated by agents in an isolated environment.
• Browser Automation: Integrated headless browser support allows agents to navigate the web, interact with pages, and extract information.
• Marketplace: Discover, share, and deploy agents and tools.
• File System: Built-in file management for agents to read/write files during execution.
• Dashboard: Monitor agent performance, manage files, and view execution history.

🛠 Tech Stack

• Framework: Next.js (App Router)
• Language: TypeScript
• Database: PostgreSQL (via Prisma)
• Authentication: NextAuth.js
• AI Integration: Vercel AI SDK, OpenAI
• UI Components: Tailwind CSS, Radix UI, Lucide Icons
• Workflow Visualization: React Flow
• Code Editor: Monaco Editor

🏁 Getting Started

Prerequisites

• Node.js (v18+)
• Docker (for sandboxed environments, if applicable)
• PostgreSQL database

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/agentos.git
   cd agentos

2. Install dependencies:

   npm install

3. Environment Setup:

   Copy the example environment file and configure your credentials:

   cp .env.example .env

   Update .env with your database URL, OpenAI API key, and NextAuth secret.

4. Database Migration:

   npx prisma migrate dev

5. Run Development Server:

   npm run dev

   Open http://localhost:3000 with your browser to see the result.

📂 Project Structure

• src/app: Application routes and pages (App Router).
  • agent/: Agent interaction and details.
  • dashboard/: User dashboard (files, profile, etc.).
  • marketplace/: Agent/Tool marketplace.
  • workbench/: Agent creation and testing workbench.
  • api/: Backend API routes (auth, execution, file system).
• src/lib: Core libraries and utilities.
  • ai/: AI-related logic (tool calling, SOP integration).
  • sop/: Standard Operating Procedure engine implementation.
  • execution/: Sandboxed code execution logic.
  • infra/: Infrastructure helpers (Prisma, Auth).
• src/components: Reusable UI components.
• doc/: Documentation and architectural plans.

📄 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.