Prerequisites for Using AI Agents for Beginners: Complete Setup Guide

To run the AI Agents for beginners course, you need Python 3.12+, an Azure subscription with Azure CLI installed, and a Microsoft Foundry project with a deployed model like GPT-4o.

The AI Agents for beginners repository from Microsoft is a hands-on course built around Jupyter notebooks that demonstrate agentic AI patterns using the Microsoft Agent Framework (MAF) and Azure AI Foundry Agent Service V2. This guide covers every prerequisite you need to configure before running the first notebook.

Core Runtime Requirements

Python 3.12 or Newer

The course relies on modern Python features and pinned package versions. In 00-course-setup/README.md (lines 94-100), Microsoft explicitly requires Python 3.12+ to ensure compatibility with the agent-framework and azure-identity libraries.

# Verify your Python version

python --version  # Should show 3.12.x or higher

Optional .NET 10 for C# Samples

If you plan to run the C#-based examples, you'll need .NET 10 or newer as noted in 00-course-setup/README.md (lines 19-24).

Azure Cloud Prerequisites

Azure CLI and Active Subscription

The notebooks authenticate using AzureCliCredential, which requires:

1. Azure CLI installed – Used for authentication without storing secrets in code (00-course-setup/README.md, lines 25-27)
2. Active Azure subscription – Required to provision Foundry resources and run the agent service (lines 26-28)

# Install Azure CLI (macOS example)

brew install azure-cli

# Authenticate

az login

# Verify active subscription

az account show

Microsoft Foundry Project with Deployed Model

In 00-course-setup/README.md (lines 45-55), Microsoft specifies that you must create:

• A Microsoft Foundry hub and project
• A deployed model (recommended: gpt-4o)

You will copy two values from this setup into your .env file:

• Project endpoint – The URL for your Foundry project
• Deployment name – The name of your deployed model

Environment Configuration

Required .env Variables

Every notebook loads configuration from a .env file. Copy .env.example to create your own:

cp .env.example .env

The minimum required variables are:

Variable	Source	Purpose
AZURE_AI_PROJECT_ENDPOINT	Foundry project overview page	Tells MAF where to send requests
AZURE_AI_MODEL_DEPLOYMENT_NAME	Model deployment list	Identifies which model to use

Optional Service-Specific Variables

Depending on which lessons you run, you may need additional configuration (00-course-setup/README.md):

• Lesson 5 (RAG): AZURE_SEARCH_SERVICE_ENDPOINT, AZURE_SEARCH_API_KEY (lines 30-38)
• Lessons 6 & 8: GITHUB_TOKEN, GITHUB_ENDPOINT, GITHUB_MODEL_ID for GitHub Models (lines 39-47)
• Lesson 8 (Grounding): BING_CONNECTION_ID for Bing search integration (lines 65-71)

Development Environment Setup

Virtual Environment and Dependencies

Microsoft recommends isolating Python packages to ensure reproducible notebook execution (00-course-setup/README.md, lines 31-33):

# Create virtual environment

python -m venv .venv

# Activate (macOS/Linux)

source .venv/bin/activate

# Activate (Windows)

.venv\Scripts\activate

# Install pinned dependencies

pip install -r requirements.txt

Efficient Repository Cloning

The repository includes ~3GB of translated notebooks. To avoid downloading unnecessary files, use a sparse or shallow clone (00-course-setup/README.md, lines 21-38):

# Shallow clone (latest commit only)

git clone --depth 1 https://github.com/microsoft/ai-agents-for-beginners.git

# Or sparse clone (English only)

git clone --filter=blob:none --sparse https://github.com/microsoft/ai-agents-for-beginners.git
cd ai-agents-for-beginners
git sparse-checkout set 00-course-setup 01-intro-to-ai-agents 02-explore-agentic-frameworks

Bootstrap Code Pattern

Every lesson notebook follows the same initialization pattern. This snippet from 01-intro-to-ai-agents/code_samples/01-python-agent-framework.ipynb demonstrates how the prerequisites connect at runtime:

import os
from dotenv import load_dotenv
from azure.identity import AzureCliCredential
from agent_framework import AzureAIProjectAgentProvider

# Load .env (created from .env.example)

load_dotenv()

# Required variables – will raise if missing

project_endpoint = os.getenv("AZURE_AI_PROJECT_ENDPOINT")
deployment_name = os.getenv("AZURE_AI_MODEL_DEPLOYMENT_NAME")

credential = AzureCliCredential()  # uses `az login` session

# Create the MAF provider that backs all agent calls

provider = AzureAIProjectAgentProvider(
    endpoint=project_endpoint,
    deployment_name=deployment_name,
    credential=credential,
)

# Example: instantiate a simple agent

from agent_framework import Agent

agent = Agent(provider=provider, name="starter-agent")
print(agent.get_capabilities())

This bootstrap pattern:

• load_dotenv() reads your .env configuration
• AzureCliCredential leverages your az login session for secure authentication
• AzureAIProjectAgentProvider establishes the connection to your Foundry project

Key Setup Files Reference

File	Purpose	Location
README.md	Course overview, lesson table, high-level prerequisites	Repository root
00-course-setup/README.md	Detailed installation steps for all prerequisites	00-course-setup/
.env.example	Template with all required and optional environment variables	Repository root
requirements.txt	Pinned Python dependencies	Repository root
01-python-agent-framework.ipynb	First working example of the bootstrap pattern	01-intro-to-ai-agents/code_samples/

Summary

• Python 3.12+ is mandatory; .NET 10+ is optional for C# samples

• Azure CLI and an active Azure subscription enable authentication and resource provisioning

• A Microsoft Foundry project with a deployed GPT-4o model provides the LLM backend

• Two environment variables (AZURE_AI_PROJECT_ENDPOINT, AZURE_AI_MODEL_DEPLOYMENT_NAME) are the minimum required configuration

• Optional services extend capabilities for specific lessons: Azure AI Search, GitHub Models, MiniMax, and Bing grounding

• Use sparse or shallow git clones to avoid downloading 3GB of translation files

Frequently Asked Questions

Can I run the notebooks without an Azure subscription?

No. The AI Agents for beginners course requires an Azure subscription to provision a Microsoft Foundry project and access the Azure AI Foundry Agent Service V2. According to the setup documentation in 00-course-setup/README.md, the AzureAIProjectAgentProvider class connects exclusively to Azure-hosted endpoints.

What is the minimum Python version required?

Python 3.12 is the minimum version required. The requirements.txt file pins dependencies that rely on Python 3.12+ features, and some notebooks use syntax introduced in recent Python versions. The setup guide explicitly warns against using older Python versions in 00-course-setup/README.md (lines 94-100).

Do I need to install all optional services before starting?

No. You only need the core prerequisites (Python, Azure CLI, Foundry project, and the two required environment variables) to run the first lessons. Optional services like Azure AI Search, GitHub Models, and Bing grounding are only required for specific lessons as noted in 00-course-setup/README.md. You can add these configurations progressively as you advance through the course.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}