A sophisticated multi-agent AI system powered by reinforcement learning and GPT-4, designed for complex task execution and continuous learning.

• Multi-Agent Architecture

  • Strategic Planning Agent
  • Research & Information Agent
  • Calculation & Analysis Agent
  • Formatting & Presentation Agent

• Advanced Capabilities

  • Neural Search with Exa AI
  • Reinforcement Learning with Experience Replay
  • Persistent Memory Management
  • Dynamic Policy Optimization
  • Context-Aware Decision Making

• Performance & Learning

  • Continuous Performance Monitoring
  • Strategy Effectiveness Analysis
  • Automated Policy Refinement
  • Resource Usage Optimization

1. Setup Environment

   # Clone repository
   git clone https://github.com/yourusername/agenta.git
   cd agenta
   
   # Create virtual environment
   python -m venv agenta
   source agenta/bin/activate  # Linux/Mac
   # or
   .\agenta\Scripts\activate  # Windows
   
   # Install dependencies
   pip install -r requirements.txt

2. Configure API Keys

   cp .env.template .env

   Add your API keys to .env:

   OPENAI_API_KEY=your_openai_key
   EXA_API_KEY=your_exa_key
   MEM0_API_KEY=your_mem0_key
   

3. Run the Agent

   python cli.py run "your objective here"

# Basic task execution
python cli.py run "analyze the latest market trends for AI companies"

# Verbose output for detailed insights
python cli.py run "calculate ROI for Project X" --verbose

# View system configuration
python cli.py info

agenta/
├── agent.py          # Core agent implementation
├── crew_agents.py    # Specialized agent definitions
├── memory_manager.py # Memory and learning systems
├── formatting.py     # Output formatting
└── cli.py           # Command-line interface

• Experience replay buffer for reinforcement learning
• Policy management with confidence scoring
• Performance metrics tracking
• Strategy pruning for optimization

• Neural search with semantic understanding
• Context-aware decision making
• Dynamic agent selection and handoff
• Continuous learning from experiences

Contributions are welcome! Please check our Contributing Guidelines for details.

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

• Documentation
• Issue Tracker
• Change Log

Built with ⚡ using OpenAI, Exa AI, and Mem0

• Code Docs RAG Search (for documentation)
• Code Interpreter (for testing/running code)
• Github Search (for code examples/solutions)
• Directory RAG Search (for codebase navigation)
• File Read/Write (for code manipulation)
• Memory Manager (for storing and retrieving memories)
• Reward Manager (for storing and retrieving rewards)
• Q-Learning (for decision making)
• Epsilon-Greedy Exploration (for exploration strategy)
• LLM (for reasoning and decision making)