Your First Conversation with Claude Code

Start Claude Code, walk through the setup screens, and ask it to do something useful with your files

Time to actually use it

Everything so far has been preparation. You opened the terminal. You installed Claude Code. You connected your account.

Now you put it to work.

If this is your very first time launching Claude Code, you'll see the setup screens covered on the previous page: theme selection, browser sign-in, security notice, and folder trust prompt. Walk through those, and you'll land on the main interface.

If you already completed that setup during installation, Claude Code opens directly.

What you're looking at

The interface is minimal.

At the bottom of the screen, there's a text input area. This is where you type your instructions, in plain English, the same way you'd message a colleague.

You might notice a grayed-out suggestion in the input area. Claude Code sometimes suggests a starting prompt based on the files it sees. Press Tab to accept the suggestion, or just start typing your own request.

That's the whole interface. A place to type, and Claude Code's responses above it. No buttons, no menus, no toolbar to learn.

Your first request

Start with something read-only — a request that looks at your files without changing anything. That way you can see how Claude Code responds before letting it modify anything.

Type this (or adapt it to whatever folder you're in):

list all the files in this folder and tell me what types they are

Press Enter.

Claude Code reads the contents of your folder and responds with a summary. Depending on what's in your folder, you might see something like:

Here are the files in this folder:

quarterly-report.docx — Word document

budget-2026.xlsx — Excel spreadsheet

team-photo.jpg — JPEG image

meeting-notes.txt — Plain text file

logo-final-v2.png — PNG image

You have 5 files: 2 documents, 2 images, and 1 text file.

Your results will look different based on your actual files, but the format is similar: file names, types, and a summary.

That's it. You typed a sentence in plain English. Claude Code read your files and answered. No commands to memorize. No special format.

When Claude Code asks permission

Now try something that involves a change. Type this:

rename all the files in this folder to use lowercase letters and replace spaces with hyphens

Before Claude Code touches anything, it pauses and asks for your permission. You'll see the specific action it wants to take — the exact file it wants to rename, the exact command it plans to run. Nothing happens until you say yes.

You get three choices:

Allow — let it do this one action
Allow, and don't ask again — let it do this type of action going forward (for the rest of this session for file edits, or permanently for this project's commands)
Deny — stop it from doing this action

For now, choose Allow each time so you can watch exactly what Claude Code does at every step.

Use your arrow keys to select your choice, then press Enter.

Tip: This permission system is your safety net. Claude Code never silently changes your files. Every edit, every rename, every deletion — it asks first.

Reading the results

After you approve, Claude Code carries out the action and shows you what it did.

For the rename example, you might see:

Quarterly Report.docx → quarterly-report.docx

Budget 2026.xlsx → budget-2026.xlsx

Team Photo.jpg → team-photo.jpg

Go check. Open the folder in Finder (Mac) or File Explorer (Windows) and confirm the files have their new names. Building the habit of verifying results matters more right now than anything else on this page.

Asking a follow-up

You're still in the same conversation, so Claude Code remembers what just happened. You can give follow-up instructions without repeating context.

Try:

now organize these files into subfolders by type — documents in one folder, images in another

Claude Code shows you its plan and asks permission before moving anything. Review the plan, approve it, and watch the files get sorted.

This is the core rhythm: you describe what you want, Claude Code figures out how to do it, you check the plan, you confirm. Ask, review, approve, verify.

When something goes wrong

Maybe Claude Code misunderstood your request. Maybe it renamed a file in a way you didn't intend. Maybe you just changed your mind.

Press Escape twice.

Claude Code snapshots your files before making changes. Pressing Escape twice rolls back both the conversation and the file changes to before the last action. You can then rephrase your request and try again.

Plain language: Escape twice is your undo button. If Claude Code does something you don't like, press Escape twice and it's as if it never happened.

Ending the conversation

When you're done, you have a few options:

Type /clear to wipe the conversation and start fresh (Claude Code stays open)
Press Ctrl+C twice to exit Claude Code entirely
Press Ctrl+D to exit

Next time you open Claude Code in the same folder, it offers to resume your last conversation or start a new one. Starting fresh each time is fine while you're learning.

What you just did

You started Claude Code. You asked it a question about your files and got a clear answer. You renamed files and reviewed every step before it happened. You organized files into folders with a single sentence. And you learned the undo shortcut in case you need to back out.

A few pages ago, you'd never opened the terminal. Now you're having a conversation with an AI that reads and rearranges the files on your computer.

Everything from here builds on the same pattern. Next up: the feedback loop, where you learn to iterate when the first result isn't quite what you wanted.