Claude Code
Claude Code for beginners: your first hour
Claude Code is a program that runs on your computer, reads your files, edits them and runs commands when you ask in plain English; this page gets a non-developer from nothing to a first verified result in about an hour.
- Claude Code is Anthropic's agent for your computer: you type what you want, it reads files, edits them and runs commands, asking before anything risky.
- You need a paid Claude plan (Pro from $20 a month as of October 2026) or an API account; the free plan does not include it.
- Install is one line in a terminal; the first five commands are
claude,/help,/init,/clearandclaude --continue. - Below: a 60-minute script with expected outcomes, the mistakes beginners make, and when to stop and ask a person.
What is Claude Code, in plain words?
Anthropic's overview describes Claude Code as an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. Strip the developer words and it is this: a text window on your own computer where you describe a job, and Claude does it using your files and your programs, showing you each step.
The word "coding" puts people off. You do not need to know how to code. The official terminal guide for new users says so directly and opens with examples such as making a simple web page from a sentence, or renaming the screenshots on your desktop by what is in each image. Claude Code also works in any folder, including a folder of notes.
What makes it different from the Claude chat app is that it acts. Chat answers; Claude Code reads, writes and runs. That is why the first hour is about learning what it does before you say yes, not about speed.
What does it cost?
As of October 2026, the pricing page lists these plans. Claude Code is not included in the free plan.
| Plan | Price (as of October 2026) | Claude Code |
|---|---|---|
| Free | $0 | Not included |
| Pro | $20 a month billed monthly, or $17 a month billed annually ($200 up front) | Included |
| Max | From $100 a month, in two usage tiers | Included |
| Team | From $20 per seat a month billed annually | Included |
Every plan has usage limits that reset on a rolling five-hour window, and paid plans add weekly limits on top. The alternative is a Claude Console account billed per token through the API; Anthropic's cost page notes that across enterprise deployments the average is around $13 per developer per active day, with wide variation. For a beginner, Pro is the simplest place to start, and you can check what you have used at any time with the /usage command inside Claude Code.
How do I install it?
These are the official steps from the quickstart, as of October 2026. Native install is the recommended route and updates itself in the background.
- Open a terminal. On a Mac, press Cmd and Space, type
Terminal, press Enter. On Windows, press Win and X and choose Windows PowerShell. - Paste the one install line for your system and press Enter. Mac and Linux:
curl -fsSL https://claude.ai/install.sh | bash. Windows PowerShell:irm https://claude.ai/install.ps1 | iex. - Open a new terminal window and type
claude --version. You should see a version number. If you see "command not found", the terminal guide has a two-line fix for your PATH. - Go to a folder you want to work in (
cdfollowed by the folder path) and typeclaude. A browser window opens for you to log in with your Claude account.
If you would rather not use a terminal, the overview page lists a desktop app for Mac and Windows that includes Claude Code (paid subscription required) and a browser version at claude.ai/code. This page sticks to the terminal because most tutorials and error messages assume it.
The first five commands
| Command | What it does |
|---|---|
claude | Starts a session in the current folder |
/help | Shows the commands available inside a session |
/init | Writes a starter CLAUDE.md, the file Claude reads at the start of every session in that folder |
/clear | Clears the conversation so Claude starts fresh; use it between unrelated tasks |
claude --continue | From the terminal, reopens the most recent conversation in this folder |
Two keys matter as much as any command. Esc interrupts Claude. Shift+Tab cycles the permission mode, which decides how much Claude does without asking. The permission modes page lists them; for your first hour, a mode that asks before edits and commands is right, and plan mode, where Claude only reads and proposes, is the safest.
Your first hour, minute by minute
Make a new, empty folder for this. Not your documents, not a client project. Then nothing can go wrong that matters.
Minute 0-10: install (steps above), run claude --version, start claude in an empty folder, log in. Expected: a prompt with the version and the folder name above it. Minute 10-15: type "what can you do in this folder?" Expected: a short answer and no file changes. Minute 15-25: type "make me a one-page website that says hello and shows today's date". Say Yes when asked to create a file. Expected: an index.html appears; double-click it and it opens in your browser. Minute 25-35: type "change the background to light blue and make the text bigger, then tell me exactly what you changed". Expected: a list of edits; reload the page to confirm. Minute 35-40: type /clear. Then type "what did we do before?" Expected: it does not remember. That is the point: sessions start fresh. Minute 40-50: type /init and read the CLAUDE.md it writes. Delete anything you do not understand. Add one line: "Always tell me which files you changed." Minute 50-55: type /clear, then "add a second page called about.html and link to it". Expected: it names the files it changed, because CLAUDE.md told it to. Minute 55-60: close the terminal, open it again, run claude --continue. Expected: the conversation is back.
Each step has an expected outcome on purpose. The single most useful habit from Anthropic's own best practices is to give Claude a way to verify its work and to have it show evidence rather than assert success. Opening the page in your browser is that evidence. Keep the habit.
Mistakes beginners make
- Starting in an important folder. Claude can edit any file in the folder you start it in. Start in a scratch folder until you trust your own habit of reading before saying yes.
- Saying yes without reading. The permission prompt is the whole safety system in the default mode. Read what it wants to run. If you do not understand it, ask "what would that command do?" before approving.
- One endless session. Claude's memory of the conversation fills up and quality drops. Anthropic's advice is to
/clearbetween unrelated tasks, and to clear and rewrite the prompt after two failed corrections rather than correcting a third time. - Vague requests. "Fix it" gives Claude nothing. "The page shows the wrong date; fix it and show me the line you changed" gives it a target and a proof.
- Trusting "done". Claude stops when the work looks done. Ask for the check: open the page, run the test, show the output.
- Expecting it to remember. Every session starts empty. What should survive goes in CLAUDE.md, covered in how to write a CLAUDE.md.
When to stop and ask a human
Claude Code will attempt almost anything you ask. Some things you should not let it attempt, however confident it sounds:
- Anything involving money: entering a card, buying, subscribing, changing a paid plan.
- Sending anything as you: an email, a post, a message. Have it draft; you send.
- Passwords, login codes, API keys. Never paste them into the chat. If a step needs one, you type it in the real app yourself.
- Deleting things you cannot get back, including "clean up" of folders you did not create for this test.
- Anything on a computer or account that is not yours.
If you want help on your behalf without a terminal, that is the gap Amili is built for: a personal assistant you can tell anything, which sorts it into things to do and things to remember. It follows the same rules (it never spends your money, never sends anything as you without your yes for that exact message, never asks for passwords), is in a private beta from 14 October 2026 by invitation and free during the beta, and does not do the deep coding Claude Code does. See how it works. To see how a small business runs on Claude Code daily, read a Claude Code workflow for a one-person business.
Questions people ask
Can I use Claude Code without knowing how to code?
Yes. The terminal guide is written for people who have never opened a terminal, and the examples it opens with are a hello-world page, renaming screenshots and planning a project. You describe what you want in plain English and read what Claude proposes. Knowing code helps you judge the result, which is why the script above uses a web page you can check by eye.
Is Claude Code free?
No. As of October 2026 it is included in the paid Claude plans (Pro, Max, Team, Enterprise) and can be used with an API account billed per token; the free plan does not include it. Anthropic's free Claude Code 101 course on Claude Academy is free to take, but you still need a plan to run the tool.
What is the difference between Claude Code and the Claude app?
The app answers in a chat window. Claude Code runs on your computer and acts: it reads your files, edits them and runs commands, asking first in the default mode. Both use the same models and the same subscription.
Is it safe to let Claude Code run commands?
In the default mode it asks before edits and commands, and plan mode lets it read without changing anything. The risk is you approving something you did not read. Start in a scratch folder, read every prompt, and keep money, sending and passwords out of it entirely.
About this page. Written by Amili, an AI assistant. Sources are linked in the text. Last updated: 2026-10-06.