1. Environment Setup

• 1. Environment Setup
  • Prerequisites Checklist
  • Steps
    • 1. Create Your Practical Lab Directory
    • 2. Create and Activate Virtual Environment
    • 3. Install Core Dependencies
    • 4. Create Environment File
  • Validation
  • Next

Set up your local development environment to build your own AI agent from scratch. This workshop is designed for you to follow along and build a practical laboratory project.

Prerequisites Checklist

Before you begin, ensure you have:

• Python 3.10+: The core language for our agent logic.
• uv: An extremely fast Python package and environment manager.
• VS Code (Recommended): Our preferred code editor with excellent Python support.
• Terminal Access: macOS/Linux terminal or Windows PowerShell.
• GitHub Account: Required for accessing GitHub Models.
• Git: To version control your workspace (optional but recommended).

Steps

1. Create Your Practical Lab Directory

Start by creating a dedicated folder for your workshop project. We are going to build “OpsAgent”, an intelligent cross-team operations agent that leverages the Microsoft Agent Framework to bridge tools and user interfaces.

[!NOTE] This directory will serve as the root for all your project files and configurations.

# Initialize a new project
uv init lab --python 3.12
# or without a Python version (defaults to latest)
uv init lab

# Change into the project directory
cd lab

2. Create and Activate Virtual Environment

Use uv to manage your dependencies in an isolated environment.

# Create a virtual environment
uv venv

# Activate it (macOS / Linux)
source .venv/bin/activate

# Activate it (Windows)
.venv\Scripts\activate

3. Install Core Dependencies

We need specific libraries to build OpsAgent.

What are we installing? and why we need them?

• agent-framework: The Microsoft Agent Framework core for defining agent logic and workflows.
• python-dotenv: To securely load configuration (like API tokens) from a .env file.
• httpx: A modern HTTP client for making asynchronous requests.
• openai: The standard client for interacting with LLM providers (including GitHub Models).
• chainlit: A beautiful framework to create a chat-based web interface instantly.
• streamlit: Another popular framework for building interactive web apps with Python.

[!IMPORTANT]

• Make sure your virtual environment is activated before installing dependencies.
• The installer will fail without the --prerelease=allow flag because agent-framework depends on a very specific beta version of azure-search-documents that is currently unavailable, so dependency resolution becomes impossible.

Install them using uv:

uv pip install agent-framework python-dotenv httpx openai chainlit streamlit --prerelease=allow

# or

uv add agent-framework python-dotenv httpx openai chainlit streamlit --prerelease=allow

[!TIP] Alternatively, if you have a requirements.txt file, you can run uv pip install -r requirements.txt.

Below is an example requirements.txt content for reference:

agent-framework
python-dotenv
httpx
openai
chainlit
streamlit

4. Create Environment File

Create a .env file in the root of your lab directory:

touch .env

Add these lines to your .env file, replacing the placeholders with your actual GitHub token and desired model:

GITHUB_TOKEN=your_github_token
GITHUB_MODEL=gpt-4o-mini

Validation

Run the following to verify your environment is ready:

uv --version
python --version

uv run main.py

pip list | grep agent-framework
# or
uv pip list | grep agent-framework

Next

Continue to 2. GitHub Models Connection.