Claude Code Basics — Mini Lesson

Claude Code is a coding assistant that lives in your terminal. You talk to it in plain English; it reads/writes files, runs commands, and builds things for you.

Starting a session

Open your Ubuntu (WSL) terminal, navigate to your project folder, type:

claude

You're now in a conversation. Type like you're talking to a person — no special syntax needed for normal requests.

The basics: just talk to it

Build me a simple app that shows a counter with a plus and minus button.

That's a complete, valid instruction. Claude will ask questions if it needs more detail, then build it.

Slash commands

Commands that start with / do something specific instead of starting a normal conversation:

| Command | What it does | |---|---| | /help | Shows what Claude Code can do | | /clear | Wipes the current conversation, starts fresh (use this between unrelated tasks) | | /init | This system's version scaffolds a new Expo project (see 04-workflow-course-overview.md, Module 2) |

Type / alone in the prompt to see the full list of commands available in your setup.

Skills

Skills are pre-built expertise packs Claude can use — for example, the expo-router skill teaches it this system's specific Expo/React Native conventions. You don't need to invoke skills by name most of the time — Claude picks the right one automatically based on what you ask. If you ever see a skill mentioned by name in a response, that's Claude telling you which expertise it's drawing on, not something you need to type yourself.

Reading Claude's responses

Tool use (file edits, commands run) shows up as it happens — you can watch it work.
When Claude finishes a big task, it tells you what changed and often suggests a next step.
If something looks wrong, say so directly: "that button doesn't work, fix it" is a completely normal, useful message.

Good habits

One task at a time. Don't ask for 5 unrelated things in one message — you'll lose track of what changed.
Test what it builds. Scan the QR code with Expo Go and try it yourself before assuming it's done. "Trust but verify."
Be specific when something's wrong. "It's broken" tells Claude nothing. "Tapping the plus button does nothing" tells it exactly where to look.
Use /clear between unrelated features. Keeps Claude focused and responses faster.

Try it now

Open your terminal in any empty folder and run:

claude

Then type:

Create a simple text file called hello.txt that says "I'm learning Claude Code" and show me the contents.

Watch what happens. That's the whole loop — ask, watch, verify.