AI-Powered Agents for OpenShift and Virtualization

Overview

Welcome to this collection of Python notebooks and tools designed to build AI-powered agents that interact with OpenShift and virtualization environments. These agents can:

• List virtual machines (VMs)
• Create migration plans
• Interact with OpenShift
• Leverage Large Language Models (LLMs) for decision-making

The notebooks demonstrate how to implement AI agents using models like Llama 3.1 and orchestrate workflows using LangChain and LangGraph. They integrate with tools such as OpenShift, virtualization platforms, and use SQLite for state management.

---

Prerequisites

To get the most out of these notebooks, please ensure you have the following installed and set up:

• Ollama: A local LLM server for running language models. Download it from ollama.com/download.
• LLM Models: We use models like llama3.1:latest and llama3.1:8b-instruct-fp16. You can download these models directly from Meta or using Ollama.

Instructions

• Downloading Ollama: Download Ollama and follow the installation instructions.

• Models We Use:

  • llama3.1:latest
  • llama3.1:8b-instruct-fp16

• Downloading Models:

  • Direct Download from Meta: Visit llama.com/llama-downloads to download models directly.
  • Using Ollama: You can download models using Ollama with the command:

  ollama pull llama3.1:latest
  ollama pull llama3.1:8b-instruct-fp16

---

Modules

Module 1: Introduction to LLMs and ReAct Prompting

1. 11-llm.ipynb - Introduction to LLMs (Llama 3.1)

Learn about Large Language Models using Llama 3.1. Understand how LLMs can be used for tasks like question answering and text generation.

2. 12-tools.ipynb - Introduction to Tool Calling

Discover how LLMs can be extended using tools to solve tasks requiring real-time information or specialized capabilities.

3. 13-react-prompting.ipynb - Introduction to ReAct Prompting

Explore the ReAct (Reasoning + Acting) prompting framework. See how ReAct enables models to reason through problems, take actions, and adjust based on observations, creating a dynamic problem-solving loop.

Module 2: Agent-Based Interactions

1. 21-agent.ipynb - Initial Agent Setup

Set up an AI-powered agent. Learn how to initialize the agent, load configurations, and connect to external services like OpenShift and the language model.

2. 22-react-agent.ipynb - ReAct Agent with Llama 3.1

Introduce the ReAct agent powered by Llama 3.1. See how the agent can execute multiple tasks, make decisions, and provide workflow feedback.

3. 23-multi-agent.ipynb - Multi-Agent Orchestration using LangChain

Learn how to orchestrate multiple agents using LangChain, allowing them to collaborate to achieve complex goals.

Module 3: Advanced Agent Applications

1. 31-planning.ipynb - AI Agent Task Planning

Explore how AI agents can plan tasks using structured processes and tools, focusing on task breakdown and execution.

2. 32-virt-agent.ipynb - Virtualization Agent

See how agents interact with virtualization platforms like VMware vSphere. Agents can list VMs, retrieve VM details, and create migration plans.

3. 33-ocp-agent.ipynb - OpenShift Agent

Discover how AI agents interact with OpenShift to manage resources such as pods, deployments, and nodes. Learn to use the OpenShift agent for handling workflows.

Module 4: Migration Workflows

1. 41-migration.ipynb - Migration Multi-Agent Workflow

Dive into creating a powerful migration workflow agent that seamlessly integrates virtualization environments with OpenShift.

---

Code and Configurations

The following directories contain code and configurations for the AI agents, services, state management, and utilities:

• agent
• prompt
• schemas
• services
• state
• utils

---

Contributing

Contributions are welcome! If you have suggestions or improvements, please open an issue or submit a pull request.

---

License

This project is licensed under the MIT License.

---

Acknowledgments

Special thanks to the developers and community behind LangChain, LangGraph, and LLM models.

---

Contact

For questions or comments, please reach out via email: [email protected]

---

Let me know if this looks good to you! I'm happy to make further adjustments.