Getting started
Most tools ask you to learn them before they’re useful. Claude Code is the other way around: it’s useful in the first ten minutes, and the learning is mostly about trust — how much rope to give it, how to see what it did, how to undo it when it’s wrong. So this chapter doesn’t start with a tour of features. It starts with you getting a real change shipped, and picks up the judgment along the way.
The audience for this chapter is the engineer who already lives in an editor — VS Code, Cursor, Copilot humming in the gutter — and is eyeing a terminal agent with reasonable suspicion. You don’t have to give any of that up. By the end of the hour Claude Code is sitting next to your editor, not replacing it, and you’ve used it to do actual work.
The hour we’ll follow
Section titled “The hour we’ll follow”You’ve just cloned a small internal service — a plain HTTP API, nothing exotic — and your onboarding buddy hands you a warm-up ticket: add a health-check endpoint and a test for it. Small, real, and verifiable. We’ll go from zero to that change shipped, in five moves:
- Install it and open it in the repo — Install Claude Code and open it in your project
- Sign in and ask it to explain the code before it touches anything — Sign in and run your first prompt
- Hand it the ticket and watch the read→edit→verify rhythm — Make your first real change
- Check its work — read the diff, undo if wrong, commit when right — Review, undo, and commit
- Wire it into your editor so this becomes a daily habit, not a detour — Run Claude Code alongside your editor
None of it is hard. The point of doing it as one continuous task is that the habits — explain before you edit, review before you trust, commit as your floor — are the same habits every later chapter builds on.