Autobot

Autobot is an autonomous AI agent designed to run on a user's desktop. It takes high-level goals from a user via a chat interface and works continuously to achieve them by planning, executing tasks, learning from outcomes, and adapting its strategy.

Features

• Autonomy: Operates continuously with minimal human intervention after initial goal setting
• Complex Task Handling: Breaks down large, complex goals into manageable steps
• Adaptive Planning: Dynamically creates, updates, and manages hierarchical plans
• Rich Tool Integration: Seamlessly utilizes a wide range of tools:
  • CLI execution
  • Python code execution
  • Computer Vision for screen analysis and interaction
  • Web Search
  • File operations
• Learning & Reflection: Analyzes successes and failures to improve future performance
• Sophisticated Context Management: Maintains and utilizes relevant context
• Privacy: Ensures all sensitive data is stored and processed locally
• User Interaction: Provides a clear chat interface for task delegation, monitoring, and intervention

Installation

1. Clone the repository:

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

2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up environment variables: Create a .env file with the following:

   OPENAI_API_KEY=your_openai_api_key
   LLM_MODEL=gpt-4.1-mini
   

Usage

1. Start Autobot:

   python main.py --project "My Project" --goal "Build a simple web application"
   

2. Access the web interface: Open your browser and navigate to http://localhost:8000

3. Interact with Autobot through the chat interface

Commands

• /status - Get current agent status
• /plan - View current plan
• /task <task> - Execute a specific task
• /reflect - Reflect on progress
• /stop or /pause - Pause the agent
• /resume - Resume the agent
• /help - Show help message

Architecture

Autobot uses a modular, layered architecture:

• Core Orchestrator: The central "brain" running the main autonomous loop
• Planning Module: Creates and manages hierarchical task plans
• Task Execution Engine: Executes individual tasks defined in the plan
• Tool Integration Layer: Provides a standardized interface for tools
• Memory & Context Manager: Manages project-related information and constructs LLM prompts
• Reflection & Learning Module: Enables the agent to learn from its experience
• LLM Interface: Wrapper for interacting with the external LLM API
• Persistence Layer: Handles reading and writing data to the local disk
• API & UI: FastAPI backend and Vue 3 frontend for user interaction

License

MIT