Launching and Logging In for the First Time

3.1 Launch and Log In: One Word Opens the Door

The install hurdle is behind you — now for the real thing. Starting Claude Code takes just three things: step into your project folder, type one word, claude, and then log in once through your browser. Logging in is a once-in-a-lifetime chore — after that it remembers you, and next time you go straight to work. Let's walk through it step by step.

Before you start, settle one thing: which folder you'll wake it up in. Claude Code treats "the folder you're currently in" as the project it works on, so the first step is to step inside. If you don't have a project yet, use the "my-practice" folder you created in Chapter 1.

1. 1
   ✋Try it

   First, step into your project folder

   Remember the cd command from Chapter 1 — it simply means "step into a folder." Type cd, add your folder's name, and press Enter. If you created "my-practice" back in Chapter 1, use the line below.

   Windows Mac Linux
2. 2
   ✋Try it

   Type one word, claude, then press Enter

   Once you're inside the folder, type the word below and press Enter. This is the magic word that "wakes up" Claude Code.

   claude

   Copy
3. 3

   The first time you start it, it asks you to log in

   The first time you run claude, it pauses and asks you to log in. The screen lists a few login options — just pick the one that best matches who you are:

   • Claude subscription account (Pro / Max) — the most common choice. If you already chat on claude.ai, just log in with that same account.
   • Claude for Teams or Enterprise — a team or enterprise account your company set up for you. If your Claude is managed and handed to you by your company, choose this and log in with your team account.
   • Anthropic Console (API, prepaid credit) — a pay-as-you-go approach. The first time you log in this way, the system automatically creates a workspace called "Claude Code" for you in the Console, so you can keep all your spending in one place to review later.
   • (Company / enterprise users) Amazon Bedrock, Google Vertex AI, and Microsoft Foundry are also supported — most individuals never need these, so if they don't mean anything to you, just skip them.

   Want to know how it works: why is spending centralized in the Console?

   If you picked the pay-as-you-go option, "Anthropic Console (API, prepaid credit)," the system automatically opens a workspace called "Claude Code" for you in the Console. Keeping all your spending in this one workspace means that later on — checking your bill, seeing how much you've spent this month, or setting a usage limit — everything is in one place at a glance, instead of scattered where you can't find it. If you log in with a subscription account, you draw on your plan's allowance and don't use this usage-based workspace, so you can skip this part.

4. 4
   ✋Try it

   A browser pops up — click "Authorize" and you're done

   Once you've chosen a login method, it automatically opens your browser and takes you to an authorization page. On that page, click "Authorize," then switch back to your terminal window — and you're logged in. The whole thing is just one click in the browser; you don't type any commands.

   ✅ What you should see

   # Back in the terminal, you'll see roughly this welcome screen (the exact wording varies by version)
   ╭─────────────────────────────────────╮
   │  ✻ Welcome to Claude Code            │
   ╰─────────────────────────────────────╯
    Logged in successfully — you're ready to start working

   Copy

3.2 Get to Know the Screen: Don't Mix Up the Two Kinds of Commands

After logging in, you'll see Claude Code's welcome screen. It shows basic information about this work session (your session), your last few conversations, and the latest update news. Don't rush to study every word — this screen is like the home page you see when you first open an app; a quick glance to get familiar is enough.

Here are two commands you can use right away, and both are genuinely handy:

• Type /help (and press Enter) → lists every command available to you. When you can't remember what's on offer, this is always a safe first move.
• Type /resume (and press Enter) → pick up an earlier conversation and continue from where you left off.

But there's one thing beginners trip over most, so let's make it clear now and save you confusion later: there are two kinds of "commands," and they're typed in different places and look different. The table below sorts them out:

3.3 Your First Conversation

Now that you know the screen, let's say your first words. Once you're inside Claude, you just type the way you'd talk to a coworker — in any language, English or otherwise; type it out and press Enter to send. There's no special syntax to learn; ask whatever you want. Let's start with one of the most useful questions:

1. 1
   ✋Try it

   Ask it, "What does this project do?"

   Just type the sentence below and press Enter to send. Once you do, Claude goes through the files in your folder on its own and gives you a plain-language summary — you never have to paste file contents to it by hand, and this is exactly what sets it apart from an ordinary chat AI.

   What does this project do?

   Copy

   ✅ What you should see

   # After reading your files, Claude replies roughly like this (the content depends on your project)
   I took a look at this project — it's mainly… (followed by a plain-language summary)

   Copy
2. 2
   ✋Try it

   Ask a follow-up and get a feel for it

   You can keep asking follow-ups in the same conversation. Try typing this one and pressing Enter to see how it answers:

   What technologies does this project use?

   Copy

3.4 The Safety System (Don't Skip This Section)

This section is short, but please read it through — it's about whether your files stay safe. The most important sentence first: before Claude Code touches your files, it always asks you first and waits for your go-ahead. It won't quietly change things behind your back.

By default, it runs in "Ask" mode: it first lays out "here's what I plan to change" in front of you, and only acts after you approve. You have two ways to approve:

• Review each change: every time it wants to change one spot, it asks you once; you look at one and approve one. This is the clearest way to see what's happening, and the best fit for beginners.
• Accept all: approve everything at once, so it stops asking you each time for the rest of this session. Convenient, but you see less of what it's doing at each step.

3.5 Ending, Reopening, and Resuming a Conversation

Whether you've had enough for today and want to wrap up, or you want to pick up where you left off next time, it's all simple. This section covers all three — closing, reopening, and resuming — in one go.

When you want to end

To close Claude Code, there are two ways — pick whichever is easier:

• Type /exit (and press Enter).
• Or press Ctrl + C twice in a row.

When you come back to work next time

To continue next time, it's the same as starting up today: step into your folder first, then launch. If you want to pick up the conversation right where you left off, you can use the "resume" approach and save yourself from recapping everything:

1. Open the terminal and use cd to step into your project folder.
2. Type claude to start fresh;
3. or type claude -r to resume your last conversation directly (you can also go in first and then type /resume).

💡 Recommended: resume your last conversation

# Resume a previous conversation (it shows a list for you to pick from)
claude -r

# Resume your "most recent" conversation (no picking — jumps straight to the last one)
claude -c

Chapter Summary

Congratulations! You've crossed the gap from "installed" to "actually up and running." Here's what you can now do: step into your project folder, wake it up with claude, log in once, say your first words to it, understand the approval prompt when it wants to change files, and close and resume a conversation.