Your first useful task
You’ve covered the ideas. Time to do the thing.
This lesson mirrors Lesson 1.10: no new theory, just a run-through where every habit from Level 2 gets used at least once. Open a real terminal. Follow along — actually type the commands.
Pick a folder
Anything you have lying around:
- The
recipes-sitefrom Level 1 (rebuild it in 30 seconds if you deleted it). - A real project you’ve been poking at — even just a folder of notes.
- A fresh empty folder —
mkdir ~/claude-demo && cd ~/claude-demoandtoucha couple of files into existence.
Step inside and confirm you’re where you think you are:
cd ~/your-folder
pwd
ls
Then:
claude
You’re in.
Question 1 — orientation
what is this project?
Watch the turn loop you learned in Lesson 2.4: thinking → tool → output → answer. Claude reads the files and tells you what it found.
If the answer is wrong or thin — that’s expected for a tiny folder. The point right now is the mechanics, not the depth.
Question 2 — pick something specific
Ask about one file:
what does the README say, in one sentence?
Or:
are there any obvious typos in any file?
This is reading, not writing — low blast radius, per Lesson 2.7. Watch Claude open the file directly. The file stays where it lives. You don’t copy-paste anything.
Question 3 — let it want to change something
Ask for a small change:
add a line to README.md that says "Built with Claude Code."
This is where you should see the permission prompt from Lesson 2.6. Read what it’s about to do. Pick allow once. Then:
cat README.md
Confirm the change is actually there. You just completed a full agent loop: ask → Claude proposes → you approve → Claude does → you verify.
When you’re done
Exit Claude (Ctrl + C twice). Look around. The folder is real. The README is really
edited. Claude was actually there.
You have officially used Claude Code.
And it doesn’t stop at code: later, in It’s not just for code, you’ll point Claude at a folder of your own files — photos, PDFs, a spreadsheet — and watch it build you a tool on the spot.
The rest of the course — nine more levels — is about learning to use it well: how to read big codebases, how to make changes safely, how to handle bigger tasks, how to make Claude yours. None of it works without the foundation you just laid.
What’s next
Level 3: The software world. A short tour of the vocabulary you’ll keep hearing — client and server, API, database, frontend and backend — so the rest of the course doesn’t drown you in jargon.