Same Room, No Buttons: Claude Code in the Terminal

There's a version of Claude Code that looks nothing like the friendly app from the last eleven lessons. No buttons. No menus. Just a black window and a blinking cursor, waiting. For a lot of people — especially anyone whose company hands them their tools — this black window is Claude Code. It's the only version they get.

If that's you, take a breath: you haven't been given the hard version. You've been given the same Claude, in a room that happens to have no buttons on the walls. Everything you learned to do by clicking, you'll do by typing — and it's fewer words than you'd think. By the end of this lesson you'll install it, open it with a single word, and recognize every single thing inside.

Why this door exists (and who it's for)

Way back in lesson two we said Claude Code is one room with several doors, and the terminal was one of them. Here's the part we can say now that you know your way around: the terminal is the original door, and the most widely used one. It came first, it runs everywhere, and it's what most workplaces install — quietly, on everyone's machine, often without the app as an option.

So this isn't the advanced elective. For a huge number of people it's day one. If a signpost — or your IT department — sent you straight here, you're in exactly the right place: get set up in this lesson, then go back and work through the course from lesson three. Every folder, every skill, every plan you make will behave precisely as described. Nothing in those lessons assumes the app.

Getting in: find it, install once, two small moves

Opening Claude Code in the terminal is a four-step ritual, and the install only ever happens once. After that, every session is the same three small moves: open your terminal, stand in your folder, knock.

curl -fsSL https://claude.ai/install.sh | bash

irm https://claude.ai/install.ps1 | iex

cd path/to/recipe-box

claude

The cockpit, re-mapped

Here's the reassuring truth about the terminal: there's nothing new to learn, only a few things to relocate. Everything you did by clicking has a typed twin. Read this table as "the thing I already do → where it moved":

One tiny wording change to expect: the careful default mode your app calls "Ask permissions" shows up in the terminal as Manual. Same mode, same behavior — Claude checks with you before it acts — just a different label under the prompt. Shift+Tab still cycles the dial you already know.

Coming back: sessions in the terminal

Close the window and yesterday's conversation isn't gone — it's just waiting for you to come back to the same folder. Two flags cover almost everything.

To pick up the most recent conversation in this folder, right where you left it:

claude -c

To choose from a list of past conversations instead — Claude shows you a picker and you arrow to the one you want:

claude -r

(-c is short for --continue, -r for --resume — either form works.) And if you're already in a session and want to hop to an older one without closing anything, /resume does it from the inside.

This is the terminal quietly rewarding a good habit: short, focused sessions you resume, rather than one endless conversation. Close freely. It'll all be here.

Everyday commands: just type a slash

You don't memorize commands in the terminal any more than you memorized menus in the app. Type a single / and a menu appears — every command you can run, searchable, right there. That menu is the skill; the names below are just the handful worth recognizing on sight:

• /init — write a starter CLAUDE.md for this project (lesson 5, one command)
• /plugin — the plugin manager, the very same one from lesson 11
• /agents — your helpers (lesson 7)
• /resume — switch to another past session without leaving
• /context — how full is Claude's desk right now (the limited desk from lesson seven)
• /usage — how much of your plan's usage allowance you've used (typing /stats opens the same view)

Everything else you might want is one slash away. Discover it the way you'd browse an app's menus: open it and look.

What the terminal actually adds

Let's be honest about the trade, because overselling it would be a disservice: the terminal does not make Claude smarter. Same model, same skills, same answers. What it adds is flow. If you already live in a keyboard-first world, Claude Code sits right where your files and tools already are — no window to switch to, everything a few keystrokes away.

It's also the doorway to automation: because it's a command, Claude Code can be woven into scripts and pipelines that run without you sitting there. That's a real and powerful room further down the hall — we're only pointing at the door, not walking through it. When you're ready for it, it'll be waiting.

(One aside, for completeness: your code editor — VS Code and friends — is yet another door into this same room, with the same everything. We don't cover it here, but if you spend your days in an editor, know that Claude Code lives there too.)

You've opened every door

So that's the last cockpit. The terminal looked like a wall and turned out to be a door — same Claude, same folder, same dial, same skills and plugins, reached by typing instead of clicking. Nothing about your setup should feel scary anymore, because you can now name every part of it from any direction.

And that's the last thing to learn. One lesson remains, and it adds no new machinery — it's the send-off: how to turn everything you now know into a first real project of your own, with the confidence you've just earned.