M1 - Mini-Lab: GitHub Copilot CLI

Week 6 · 20 points · pass/fail · take home · individual · about 1 hour · submit in Canvas

Why you are doing this

AI coding agents that run in your terminal are now a normal part of professional software work. They read your files, propose changes, run commands, and explain what they did. Used well they save time; used carelessly they write code you do not understand and cannot debug. The only way to learn the difference is to drive one yourself on something small enough that you can check every line.

In this mini-lab you install GitHub Copilot CLI, point it at an empty directory, and have it help you build a tiny C program that does something you already know how to do by hand: resolve a hostname, the way dig did in A1. Then you ask it to change the program, ask it to explain part of it, and save the whole conversation as proof.

This is a take-home mini-lab: you do it on your own, outside of class, and submit it individually. It is much smaller than a project and does not use the project starter repository or the project grading rubric.

This costs you nothing

Every student in this course already has GitHub Copilot through the free GitHub Education benefit. You do not need to pay for anything to complete this lab. Never enter a credit card, never buy a subscription, and never buy extra requests for this course.

If Copilot ever tells you that you are out of requests or asks you to upgrade, stop and email me. That is a problem for me to solve, not you.

Before you start

1. Confirm your Copilot access. Sign in to GitHub and open github.com/settings/copilot. It should show that you have Copilot through GitHub Education. If it does not, you can sign up for a pro account at https://education.github.com/pack. You will need to use your student (edu) email address to qualify for the free stuff.
2. Pick a machine. Your own laptop is the easiest place to do this. You need a terminal and a C compiler (cc or gcc) with make.
3. Budget about an hour. Most of it is reading what Copilot produces.

Part 1 - Install and log in

Install Copilot CLI with one of the following. Pick whichever matches your machine.

Platform	Command
macOS or Linux (Homebrew)	brew install --cask copilot-cli
macOS or Linux (no Homebrew)	curl -fsSL https://gh.io/copilot-install | bash
Windows	winget install GitHub.Copilot
Any platform with Node.js 22 or later	npm install -g @github/copilot

On Windows you need PowerShell 7 or later; WSL also works, using the Linux instructions.

Then check that it runs and log in:

copilot --version
copilot

The first time it starts, it asks you to log in. Type /login at its prompt and follow the instructions, which open a browser to GitHub. Use the same GitHub account that has your Education benefit. When you are logged in, type /exit.

The official instructions are at Installing GitHub Copilot CLI if anything above does not work on your machine.

Stay on the default model

Copilot CLI lets you switch models with /model. Leave it on the default. Some models use up your monthly allowance much faster than others, and this lab does not need a fancy one.

Part 2 - Build lookup with Copilot

Make an empty directory for the lab and start Copilot inside it:

mkdir cs425-m1
cd cs425-m1
copilot

Copilot asks whether you trust the files in this folder. Say yes: it is your own, empty directory. Never say yes in a folder you do not own, such as your home directory.

Prompt 1: write the program

Give Copilot this prompt, word for word or in your own words:

Write a small C program called lookup.c that takes one hostname on the command line, resolves it with getaddrinfo, and prints every address it gets back, one per line, labeled IPv4 or IPv6. Print a useful error and exit with a nonzero status if the lookup fails. Also write a Makefile that builds it with -Wall -Wextra and has a clean target.

Copilot will propose creating files and running commands, and it asks permission for each one. Read each request before you approve it. Approve each action individually rather than choosing the option that allows everything for the rest of the session. You are responsible for every command it runs on your machine.

When it is done, build and run the program yourself, in a second terminal or after /exit:

make
./lookup example.com
./lookup www.boisestate.edu
./lookup no-such-host.invalid

Compare the addresses with what dig example.com A and dig example.com AAAA report. They should agree.

Prompt 2: change the program

Ask Copilot to add a feature:

Add -4 and -6 options to lookup so it prints only IPv4 or only IPv6 addresses. With no option, print both.

Rebuild and run ./lookup -4 example.com and ./lookup -6 example.com. If something does not compile or does the wrong thing, tell Copilot what happened and let it fix it. That back and forth is part of the lab, so leave it in the log.

Prompt 3: make it explain

Ask Copilot at least one question about the code it wrote. Some ideas:

• Why does getaddrinfo sometimes return the same address more than once?
• What does hints.ai_socktype do, and what happens if you leave it at zero?
• Why does the program need freeaddrinfo?

Check its answer against man getaddrinfo. Copilot is often right and sometimes confidently wrong; that is exactly why you check.

Part 3 - Save the conversation

While you are still in the Copilot session, export it to a Markdown file:

/share file ./copilot-session.md

Open copilot-session.md and make sure it contains your prompts and Copilot's answers. If you closed Copilot before exporting, run copilot --resume in the same directory, pick the session, and export it then.

Part 4 - Reflect

Write a short reflection.md (a few sentences each is plenty):

1. One thing Copilot got right that would have taken you longer by hand.
2. One thing you had to check, correct, or did not trust. If everything worked the first time, say what you checked and how.
3. Would you use a tool like this on a course project? What would you want to understand yourself before you did?

What to submit

Upload these four files to the Canvas assignment:

File	What it shows
copilot-session.md	The exported conversation, with all three prompts
lookup.c	The program Copilot helped you write
Makefile	The build file
reflection.md	Your answers to Part 4

Do not zip them; upload the four files directly.

Grading

This is pass/fail. You pass if:

• the session log shows you actually used Copilot CLI for all three prompts,
• lookup.c and the Makefile match what is in the log and build with make, and
• reflection.md answers all three questions in your own words.

Code that has a bug still passes, as long as the log and the reflection show you worked with it honestly.

What about Claude or other tools?

You may use Claude Code instead of Copilot CLI if you already have access to it. The steps are the same: install it, run it in an empty directory, give it the three prompts, and export the conversation with its /export command.

Claude Code is not free, and there is no advantage to using it for this lab. Do not pay for Claude, or for any other AI tool, to complete this assignment. Copilot CLI through your GitHub Education benefit is all you need, and it is what I will help you troubleshoot.