Aegis Multi-Agent Framework

A powerful, extensible platform for building and deploying AI agent systems with seamless local LLM integration.

Overview

The Aegis Multi-Agent Framework provides a robust foundation for creating sophisticated multi-agent systems while maintaining simplicity and flexibility. Perfect for both researchers and developers looking to build advanced AI agent applications.

Key Features

• Modular Agent Architecture

  • Plug-and-play agent components
  • Customizable agent behaviors
  • Extensible design patterns

• Local LLM Integration

  • Native Ollama support
  • Multiple model compatibility
  • Optimized inference pipeline

• Advanced Task Management

  • Real-time task monitoring
  • Parallel task execution
  • Priority-based scheduling

Quick Start

Installation

pip install aegis-framework

Basic Usage

from aegis_framework import MasterAIAgent, DesignAgent

# Initialize a master agent
agent = MasterAIAgent(model="gemma2:9b")

# Generate responses
response = agent.answer_question(
    "What are the key principles of multi-agent systems?"
)
print(response)

# Create a specialized design agent
designer = DesignAgent(model="gemma2:9b")
design = designer.generate_new_design(
    context="Create a microservices architecture",
    constraints=["scalability", "fault-tolerance"]
)
print(design)

Creating Custom Agents

from aegis_framework import MasterAIAgent
from typing import Dict, Any, Optional

class DataAnalysisAgent(MasterAIAgent):
    def __init__(
        self,
        model: str = "gemma2:9b",
        custom_tasks: Optional[Dict[str, List[str]]] = None
    ):
        super().__init__(model=model)
        
        # Add specialized tasks
        self.agent_task_map.update({
            "data_analysis": [
                "analyze data",
                "statistical analysis",
                "trend analysis",
                "data visualization"
            ]
        })
        
        if custom_tasks:
            self.agent_task_map.update(custom_tasks)
    
    def analyze_data(
        self,
        data: str,
        analysis_type: str = "comprehensive"
    ) -> Dict[str, Any]:
        """Perform data analysis with specified parameters."""
        prompt = f"Analyze this {analysis_type} data: {data}"
        return self.perform_task(prompt)

# Usage
analyst = DataAnalysisAgent()
results = analyst.analyze_data(
    data="your_data_here",
    analysis_type="statistical"
)

System Requirements

• Python 3.7+
• Ollama (for local LLM support)
• 8GB+ RAM (recommended)
• CUDA-compatible GPU (optional)

Example Scripts

The framework includes several example scripts to help you get started:

1. basic_usage.py: Demonstrates core functionality
2. design_agent_example.py: Shows advanced design capabilities
3. custom_agent_example.py: Illustrates custom agent creation

Run any example with the --help flag to see available options:

python examples/basic_usage.py --help

Version History

Current Version: 0.1.15

Key Updates:

• Enhanced local LLM integration
• Improved design agent capabilities
• Better error handling
• More comprehensive examples

License

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

Contact

• Author: Metis Analytics
• Email: cjohnson@metisos.com
• GitHub Issues: Report a bug

Acknowledgments

Special thanks to:

• The Ollama team for their excellent LLM runtime
• Our contributors and early adopters
• The open-source AI community

Made with ❤️ by Metis Analytics