Your assignment is to build a Thingy: an application that you think would be fun to have. Any Thingy. A game, a tool, a web app, a toy, something useful, something completely useless — it's your call.
The catch is that you don't write it yourself. You direct an agentic AI, such as Claude, to build it for you. Your job is to decide what to make, explain it clearly, look at what comes back, and steer until you like it.
The point of this lab is simple: to let you see for yourself how powerful agentic AI has become, and to break the ice — you will meet your classmates through what they build. That lesson will carry through every lab that follows.
Due: the end of the first week of classes. Everything in this lab works on Mac, Windows, and Linux, and no setup is required beyond two quick installs, so you can start today.
This is a competition that everyone wins. In class, students take turns showing their Thingy to everyone else. After each demo, the audience applauds based on how much they like it. That's the whole scoring system.
So bring something you'll enjoy showing off. The best Thingies are usually the ones their creators actually wanted to exist.
Every student who builds a Thingy, shows it, and hands in the file described in Step 5 receives 100% credit. No exceptions, no judging, no rubric. The applause is for glory only.
Markdown is plain text with a few simple formatting
conventions, saved in a file ending in .md. It is
the standard format for writing instructions to an AI assistant,
and you will use it all semester. A Markdown editor shows you
the formatted result while you type.
Follow the Markdown editor installation page to install one on your Mac, Windows, or Linux machine.
Before you talk to the AI, write down what you want. Download
the prompt template, save it as
a .md file, and fill it in with the specifics of
your Thingy: what it is, its features, how it should look, and
what "done" means.
The template begins with a Guidelines section: standing rules that apply to everything the AI produces, before any specifics of your Thingy. We have filled it in with rules for clear writing. Read them — they are good rules for your own writing too — and change them if you disagree.
By the way, this is exactly how this course is built: each lab page starts as a Markdown file given to Claude. You can read the one for this lab here: instructions.md.
Your finished prompt .md file, with the
Reflection section at its end filled in.
One more recommendation: be nice to your Claude. When the result isn't what you wanted, explain what you meant instead of just saying it's wrong. Patient, clear direction gets better work — from AIs and from people — and it makes the whole project more fun.
Install the Claude desktop app from claude.ai/download (or use another agentic AI). The ordinary app is all you need — the full command-line computing environment comes in Lab 2.
Then drop your .md file into Claude and ask it to
build what the file describes. Look at what comes back, and
steer.
Building is an iterative loop, and the .md file is
the record of where the design stands. When you change your
mind about a feature, update the file and give it to Claude
again. You can also ask Claude to update the file to match what
you have agreed on. When the Thingy matches the file and you
like it, you are done.
Bring your Thingy to class and demo it. Applause follows.
Submit one file on Brightspace: your prompt
.md file. Fill in the Reflection section
after your demo and before you submit. A few sentences under
each of its headings is plenty:
.md file, with the Reflection
section filled in.
One .md file, and nothing else. No zip files, no
PDFs, and no code. We saw the code run in class.