Get an AI coding partner running right on your server

Your server is ready and its toolbox is stocked. Now for the exciting part: bringing in an AI helper that can read your files, write real code, and run commands — right there in the terminal, working alongside you. This module gets Claude Code installed and shows you how to work with it well.

Meet your new pair programmer

Claude Code is a version of Claude that lives inside your terminal instead of a chat window in a browser. You describe what you want in plain English, and it reads your project's files, writes new ones, edits existing code, and can run commands on your server — always pausing to ask your permission before it changes anything.

Picture a patient, expert programmer sitting right next to you at the keyboard. You explain the goal in your own words. They do the typing, explain their choices as they go, and check with you before anything risky. That's the relationship you're about to build.

Two ways to run Claude Code

There are two common setups for using Claude Code on a project like this one. Both genuinely work. This course picks one on purpose, to keep things simple.

Workflow A: run it on your server (what we'll do)

You already know how to connect to your VPS. In this workflow, you connect the same way, then start Claude Code right there. It writes files directly into your project folder on the server — the exact place your site will live. Nothing needs to be copied or uploaded afterward, because it was never anywhere else. For a first course, this is the simplest path, and it's the one we'll use starting today.

Workflow B: run it on your laptop, then send the files up

The other option is to install Claude Code on your own laptop, build your project there, and then transfer the finished files to your server afterward — a step usually called "deploying." It's a completely normal way to work, and plenty of professional developers do exactly this. But it adds a moving part: two computers to keep in sync, and an upload step every time something changes. It's worth knowing this exists, but it isn't what this course focuses on.

Installing Claude Code

Claude Code needs Node.js to run — the same Node.js you installed back in Module 5. If you skipped that module, go set it up first; everything here builds on it.

With that caveat in mind, here's the usual way to install it. Connect to your server over SSH, then run:

npm install -g @anthropic-ai/claude-code

This downloads Claude Code and makes a new claude command available anywhere on your server. Next, move into the project folder you want to work in, and start it up:

claude

The first time you run it, Claude Code will walk you through signing in with your Anthropic account — the same one from Module 0. Usually, it gives you a link to open in your browser, you approve access there, and you come back to the terminal already signed in. Just follow whatever it shows you on screen; if the exact wording looks a little different from what's described here, that's normal — trust docs.claude.com over anything else, including this page.

How to talk to Claude so it actually helps

Claude Code is powerful, but like any collaborator, it works best when you communicate clearly. A few habits make a huge difference:

• State a clear goal. "Make a simple web page that says hello and shows the time" is a great instruction — specific and small.
• Go one step at a time. Let Claude finish a change and show it to you, then ask for the next one — rather than piling on five requests at once.
• Ask it to explain what it did. "Explain that in plain English" turns a wall of code into something you actually understand.
• If something breaks, paste the error back. Claude can usually read an error message and fix the problem faster than you could describe it yourself.
• Keep requests small and specific. "Build my whole website" is much harder to review than five small requests in a row.

Hands-on: build something tiny, together

Let's actually use it. This is just a warm-up — a way to see Claude Code work before Module 7, where we build the real project. It takes about five minutes.

1. Make a project folder

   Connect to your server, then create a small folder to work in:

   terminal Copy

   cd ~/apps
   mkdir hello && cd hello

2. Start Claude Code

   From inside that folder, run:

   terminal Copy

   claude

3. Ask for something tiny

   Once Claude Code is ready, type a request like this one:

   inside claude Copy

   Create a tiny web page that says Hello, world and shows today's date, and tell me how to open it.

4. Review what it built

   Claude will show you the files it wants to create before it creates them, then explain what it did. Read that explanation — it's genuinely one of the best ways to learn. Don't worry about actually opening or running the page yet; that's exactly what Modules 7 and 8 are for. Right now, just notice this: you described an idea in plain English, and real files now exist because of it.