AI Agent Hands-on Environment Setup

This module prepares three things you need to work with an AI agent on a Windows 11 laptop: install and configure the ChatGPT desktop app (Chat, ChatGPT Work, and Codex); install VS Code with the matching extensions; and register a personal GitHub account and submit a GitHub Education application. The three parts are independent — work through them at your own pace, then check off all three at the end.

Learning Objectives

After this module, participants can:

• Install and sign in to the ChatGPT desktop app, and choose Chat, ChatGPT Work, or Codex for the task at hand.
• Install the Codex and GitHub Copilot extensions in VS Code so the AI agent can read, write, and review code directly in the editor.
• Install and use syntax highlighting for Python, JavaScript, Markdown, and HTML in VS Code, plus direct previews for MS Office files (docx/xlsx/pptx), without switching to another app to view course material.
• Register a personal GitHub account and submit a GitHub Education (teacher / Global Campus) verification application.
• Use the checklist to verify all three parts are set up and troubleshoot common issues.

Prerequisites

Minimum Computer Requirements

This module only needs VS Code, the ChatGPT desktop app, and a browser open at the same time, so the requirements are lighter than a full developer toolchain. The "Recommended" column marks the threshold for a smoother experience.

8 GB of RAM is enough to complete every task in this module, but 16 GB is noticeably smoother once VS Code, a browser, and the ChatGPT desktop app are all open together.

Overview of the Three Parts

The three things this module covers are independent of each other — work through them in whatever order suits you:

Part 1: Install and Configure the ChatGPT App (Chat / Work / Codex)

The ChatGPT desktop app is the shared entry point on Windows for Chat, ChatGPT Work, and Codex.

After signing in, pick the right interface for the task:

• Use Chat for quick questions.
• Select ChatGPT, then Work, for research, documents, or a multi-step, reviewable outcome.
• Select Codex from the product menu for a course project, code, commands, and diffs.

⚠️ Before sending the first task in Codex, choose Ask for approval; don't enable Full access by default. That way you can review what the agent intends to do and what it changed at every step, before deciding whether to approve it.

Check: run the guided "Hello, AI Agent" task (ask Codex to create a hello.txt file in the project folder) to confirm the AI agent can receive instructions, produce a file, and report back.

Part 2: VS Code and Extensions

VS Code is the main window for working with the AI agent later. With the extensions below, you can handle both code collaboration and course-material review in the same window.

Install these by searching in the VS Code Extensions panel, or run them one by one in a terminal:

code --install-extension openai.chatgpt
code --install-extension GitHub.copilot
code --install-extension ms-python.python
code --install-extension yzhang.markdown-all-in-one
code --install-extension ms-vscode.live-server
code --install-extension cweijan.vscode-office

⚠️ Extension IDs change as the Marketplace is updated. Before the live session, search each name in the Extensions panel to confirm the ID is still correct — especially the Office document preview extension, which is the least mainstream of the set.

Check: open the Extensions panel and confirm the extensions above are installed and enabled; then open a sample .py, .md, .html, and a .docx or .xlsx file, and confirm syntax highlighting or the preview renders correctly.

Alternatives: More Free AI Agent Options

Beyond the ChatGPT app/Codex from Part 1, here are two more AI agent options worth knowing about — both free, with quotas that are typically more than enough for workshop-scale tasks — for anyone who wants more flexibility: Google Antigravity, and the education-tier GitHub Copilot unlocked via Part 3's GitHub Education.

Google Antigravity

Like the ChatGPT app/Codex, Google Antigravity can take instructions, read and write code, and run tasks, powered by Gemini 3.1 Pro/Flash. It's called out here because, during its current public preview, the free usage quota on a personal Google (Gmail) account is generous, with no credit card required. It also bundles a VS Code-like editor, so choosing this alternative lets you skip Parts 1 and 2 (installing the ChatGPT desktop app and VS Code separately) and go straight to Part 3 (GitHub account) once it's installed.

⚠️ Antigravity has adjustable approval levels for both terminal commands and agent-made changes (a Terminal Execution Policy of Off/Auto/Turbo, plus an Artifact Review policy). The first time you use it, pick a setting that requires manual confirmation so you can review what the agent intends to run or change before approving it — then loosen the settings once you trust its behavior.

Check: open Antigravity and confirm you can sign in with a personal Google account; under a setting that requires manual confirmation, run the equivalent "Hello, AI Agent" task (ask the agent to create a hello.txt file) and confirm the agent can receive instructions, produce a file, and report back.

GitHub Copilot (via GitHub Education)

The GitHub Copilot extension installed in Part 2 defaults to a paid plan; once your GitHub Education faculty verification in Part 3 is approved, Copilot automatically switches to the education tier — a more generous, free plan — with no extra extension to install or configure.

Check: once GitHub Education is approved, confirm the Copilot icon/account status in VS Code shows the education/free tier, and that Copilot Chat works normally for asking about or editing code.

Part 3: Apply for a GitHub Account and GitHub Education

A GitHub account is the foundation for managing projects and files in later modules; GitHub Education lets a verified teacher unlock education-tier benefits for tools like Copilot.

A school-domain email isn't required, but registering with one usually speeds up GitHub Education's automatic verification.

Check: confirm you're signed in at github.com; confirm the GitHub Education application has been submitted (the page shows "Application submitted", or already shows the approved benefits page).

Hands-on Activities

Once all three parts are set up, run through the checklist below for a final check.