AI CODING TOOLS / LESSON 03
Your First Conversation.
Talking to Claude Code for the first time
Loading the lesson…
Loading activities…
Browse lessons
AI CODING TOOLS / LESSON 03
Talking to Claude Code for the first time
Loading the lesson…
Loading activities…
Claude Code is installed. Cool. Now let's actually USE the thing! In this lesson you'll start Claude Code for the first time and have a real conversation with your AI programming assistant.
Open your terminal and head into any folder where you might want to work on a project. Then type this.
That's it. Just the word "claude" and hit Enter.
The first time you run it, Claude Code asks you to log in with your Claude account. Follow the prompts to sign in, and after that you'll get a welcome message and a spot to type.
Once Claude Code starts, you'll see something like this.
That > is your prompt, which is where you type your messages. It's basically a text conversation... except instead of texting a friend, you're texting an AI that can code.
Let's keep it simple. Type this and press Enter.
Claude replies with a friendly rundown of what it can do. Notice it writes in full sentences and explains things clearly. That's how all your conversations will go.
Now let's get Claude to actually DO something. Try this one.
Then watch what happens.
Claude reads your request and works out what you need. You might see it say something like "I'll create an HTML file for you."
It describes what it's about to do. That way you understand what's coming and can make sure you agree before anything changes.
Before it creates the file, Claude checks whether that's okay. You can say yes, say no, or ask it to change something. (Current versions usually start you in auto mode, which skips a lot of these check-ins. The permission modes lesson shows you how to switch.)
Once you approve, Claude makes the file. You'll see the code it wrote and the filename.
Claude Code answers you in a few different ways.
Ask a question and Claude explains things in plain language. It might throw in examples or break things into steps.
When Claude writes code, it shows it in a special formatted block. The code gets highlighted in colors so it's easier to read.
When Claude creates or edits files, it tells you exactly what it's doing. You'll see it name every file it's creating or editing, like index.html or style.css.
Sometimes Claude needs to run commands (like installing a package or starting a server). It shows you the command and asks permission first.
Here's some stuff to try. Type any of these and see what Claude does with it.
Ask a question.
Ask for code.
Ask for an explanation.
Ask for help.
Honestly, this part matters more than anything I could write here. Courses don't really teach you anything... the good ones just clear out the noise so you can go DO the thing. Real learning happens when you start typing.
One of the best things about Claude Code is that it remembers your conversation, so you can go back and forth with it.
Here's how that looks.
You. Create a webpage with a button
Claude. [Creates the file]
You. Now make the button green instead of the default color
Claude. [Updates the file]
You. Add another button that says "Cancel"
Claude. [Adds the second button]
This is where it gets fun. You build things one step at a time and tweak as you go.
Claude Code has some built-in commands you can use, and they all start with a forward slash.
/help - Shows all available commands/model - View or switch the AI model/context - Shows how full your context window is/status - Shows your version, model and account/clear - Clears the conversation and starts fresh/compact - Summarizes conversation to free context space/usage - Shows session cost and your plan's usage limits (/cost does the same thing)/exit or /quit - Closes Claude CodeType /help to see the full list.
When you're done, there are two ways out.
/exit and press EnterCtrl + C (hold Control and press C) twice, since the first press just clears or cancelsEither way, Claude Code closes and you're back at your regular terminal.
In just a few minutes you went from zero to this.
That's the basics of working with Claude Code. The more you use it, the more natural it feels (kinda like driving stick, minus the stalling).
claude in your terminal/help to see available commands and /exit to quit/context to keep an eye on your context window and /compact to free up spaceNext lesson we'll look at how Claude Code actually thinks and works. Once you get this "agentic loop" thing, working with it gets a whole lot easier...