AI literacy for students
Your practical control room for AI in 2026.
A bilingual field guide for the workshop: understand agents, set up the right tools, practice safe workflows, and ship a small portfolio before the room goes quiet.
Agent workflow preview
prompt → inspect → edit → verify
I want to build a portfolio with AI, but I do not know what to install first.
We will map the tools, install the basics, run OpenCode, review each diff, then deploy the result.
field notes
Built for the questions students actually ask
The guide is shaped around the confusion points that appear in the first hour: which tool to trust, what an agent can change, where secrets belong, and how to know when to stop.
Before setup
What is the difference between ChatGPT, Cursor, Claude, and OpenCode?
During practice
If the agent wants to run a command, how do I know it is safe?
After shipping
How do I keep the site online and update it without breaking it?
Before setup
What is the difference between ChatGPT, Cursor, Claude, and OpenCode?
During practice
If the agent wants to run a command, how do I know it is safe?
After shipping
How do I keep the site online and update it without breaking it?
Toolchain
Bring the tools you already recognize. Learn where each one belongs.
This is not a logo wall. Every tool below has a job in the workshop, from asking the first question to deploying the final portfolio.
ChatGPT
The familiar starting point for questions, drafts, explanations, and quick checks.
Claude
Useful for careful reasoning, code review, long edits, and comparing answers.
OpenCode
The hands-on terminal agent we install and run together during the workshop.
Cursor
An AI-native code editor, close to VS Code but with chat and agents wired in.
GitHub
Where your code lives, gets reviewed, and can publish your portfolio.
Node.js
The JavaScript runtime behind the portfolio project and modern web tools.
End state
One afternoon, three concrete wins.
A working local setup
Node, npm, an editor, GitHub basics, and OpenCode installed on your own laptop.
A real agent loop
Prompt, inspect, tool call, diff review, fix, verify. You see the loop before you trust it.
A live portfolio URL
A small personal site deployed online, with a clear path for future updates.
Lesson path
Eight short lessons, one clear arc.
The sequence moves from orientation to setup, then hands-on agent work, safety, and deployment.
- 01Welcome to AI in 2026Understand why AI changed so quickly and what students should learn first.8 min
- 02What AI agents areLearn what makes an agent different from a chatbot and where OpenCode fits.14 min
- 03OpenCode before installationPreview the agent loop before installing anything: prompt, inspect, propose, edit, verify.12 min
- 04Windows setup pathPrepare a Windows machine for practical AI tools and agent workflows.18 min
- 05Basic Linux guideLearn the small set of Linux terminal habits that make AI agent work safer.14 min
- 06Node and npmInstall the JavaScript runtime and package manager used by the portfolio project.16 min
- 07OpenCode hands-onInstall OpenCode, open a project, and practice safe agent prompts.22 min
- 08Build your portfolioUse the agent to build a small personal portfolio that can be deployed.35 min
- 09GitHub and secrets safetyPublish code without leaking API keys, tokens, or environment files.20 min
- 10Deploy your siteChoose a hosting guide and get the portfolio online.20 min
10/10 lessons ready
Safety layer
The agent does the work. You keep judgment in the loop.
Speed is only useful when the student still understands what changed.
- rule 01
Never paste API keys, tokens, or .env files into chats or screenshots.
- rule 02
Read terminal commands before you run them. Curiosity is cheaper than recovery.
- rule 03
Treat every diff as a proposal until you accept it.
Start with the map, then build the thing.
The first lesson gives students a clean mental model before the workshop moves into installs and hands-on agent work.
Open lesson one