agentic-fleet 0.3.6

A fleet of agentic AI assistants

Agentic Fleet

Agentic Fleet

A powerful framework for building and managing fleets of AI agents.

Vision

Agentic Fleet aims to provide a flexible and robust platform for developing and deploying sophisticated multi-agent systems. We strive to empower developers to create AI solutions that can tackle complex tasks through coordinated effort and intelligent collaboration.

Features

• 🤖 Multiple Agent Types: Agentic Fleet supports a variety of specialized agents, each designed with specific capabilities:

  • Planner: Decomposes complex tasks into manageable steps and creates execution plans.
  • Executor: Executes tasks, monitors progress, and handles task completion.
  • Researcher: Gathers and analyzes information from various sources to support decision-making.
  • Critic: Evaluates the quality of work, provides feedback, and ensures adherence to standards.
  • Video Surfer: Analyzes video content, extracts key information, and summarizes findings.
  • File Surfer: Navigates and manipulates file systems, enabling agents to interact with local data.
  • Web Surfer: Interacts with web content, retrieves information, and automates web-based tasks.

• 🔄 Advanced Prompt Management: Efficiently manage and optimize agent behavior with:

  • Version control for prompts, allowing for iterative refinement and rollback.
  • Hot reloading capabilities, enabling real-time updates to prompt configurations.
  • A/B testing support for comparing the effectiveness of different prompts.
  • Metrics collection and analysis to understand prompt performance and impact.

• 🛠️ Extensible Architecture: Build upon a solid foundation with:

  • Modular design that promotes flexibility and maintainability.
  • Easy agent customization, allowing developers to tailor agents to specific needs.
  • Flexible tool integration, enabling agents to leverage a wide range of functionalities.
  • Robust error handling to ensure reliable operation and graceful recovery.

• 📊 Monitoring & Analytics: Gain insights into your agent fleet's performance with:

  • Detailed metrics tracking to monitor key performance indicators.
  • Performance analysis tools to identify bottlenecks and areas for optimization.
  • Usage statistics to understand agent activity and resource consumption.
  • Error reporting to quickly identify and address issues.

Installation

Using pip

pip install agentic-fleet

Using PDM (recommended)

pdm add agentic-fleet

Development Installation

To set up a development environment:

git clone https://github.com/yourusername/agentic-fleet.git
cd agentic-fleet
pdm install

Quick Start

from fleets import PlannerAgent, ExecutorAgent, ResearcherAgent

# Initialize agents
planner = PlannerAgent()
executor = ExecutorAgent()
researcher = ResearcherAgent()

# Create a task
task = "Build a web application with user authentication"

# Plan the task
plan = await planner.create_plan(task)

# Research requirements
research = await researcher.gather_information(task)

# Execute the plan
result = await executor.execute_plan(plan)

Configuration

The framework can be configured using environment variables or a .env file:

AZURE_OPENAI_API_KEY=your_api_key
AZURE_OPENAI_ENDPOINT=your_endpoint
AZURE_OPENAI_API_VERSION=2023-12-01

Documentation

Detailed documentation is available at docs/:

• Architecture Overview
• Agent Types
• Prompt Management
• API Reference

Contributing

We welcome contributions! Please see our Contributing Guide and our Code of Conduct for details on how to contribute to Agentic Fleet.

Get Involved

• Code of Conduct: Please review our Code of Conduct to understand our community standards.

• GitHub Issues: Report bugs, suggest features, and see known issues.

• GitHub Discussions: Share ideas, ask questions, and engage with the community.

Development Setup

1. Fork the repository
2. Create a virtual environment
3. Install dependencies:

   pdm install --dev
   

4. Run tests:

   pdm run pytest
   

License

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

Acknowledgments

• Built with AutoGen
• Inspired by the work of the AI research community

Support

• 📫 Email: [e-mail]
• 🐛 Issues: GitHub Issues
• 💬 Discussions: GitHub Discussions