Claude: Zero to Hero
Claude Code · module 1 of 13

Install and first session

What you’ll learn: get Claude Code running and complete a real task with it.

Install

macOS, Linux, WSL:

curl -fsSL https://claude.ai/install.sh | bash

Windows PowerShell:

irm https://claude.ai/install.ps1 | iex

Windows CMD:

curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

If you see The token '&&' is not a valid statement separator, you’re in PowerShell, not CMD. If you see 'irm' is not recognized, you’re in CMD, not PowerShell. Your prompt shows PS C:\ in PowerShell.

Native installs auto-update in the background.

Alternatives:

brew install --cask claude-code          # stable channel, ~1 week behind
brew install --cask claude-code@latest   # latest channel
winget install Anthropic.ClaudeCode      # Windows

Homebrew and WinGet installs do not auto-update — run brew upgrade claude-code / winget upgrade Anthropic.ClaudeCode periodically. apt, dnf and apk are also supported.

On native Windows, Git for Windows is recommended so Claude Code can use the Bash tool; without it, PowerShell is used as the shell. WSL doesn’t need it.

Verify:

claude --version

Other surfaces

Surface How
VS Code / Cursor Search “Claude Code” in Extensions, or install from the marketplace
JetBrains Install the Claude Code plugin; requires the CLI installed separately
Desktop app claude.com/download, then the Code tab
Web claude.ai/code — no local setup
Mobile The Claude app for iOS and Android

All surfaces share the same engine — your CLAUDE.md, settings, and MCP servers work across every one.

Log in

claude

You’ll be prompted on first use. Options:

Later: /login inside a session to switch accounts, or from the shell:

claude auth login            # sign in
claude auth login --console  # use Console billing instead of a subscription
claude auth status           # check; exits 0 if logged in
claude auth logout

If ANTHROPIC_API_KEY is set, Claude Code skips the login prompt and asks you to approve the key.

Your first session

cd /path/to/your/project
claude

You’ll see the prompt with version, current model, and working directory above it.

Start by asking, not doing

Before you let it change anything, spend five minutes getting a feel:

what does this project do?
what technologies does this project use?
where is the main entry point?
explain the folder structure

Claude reads files as needed. You don’t have to add context manually.

You can also ask about Claude Code itself:

what can Claude Code do?
how do I create custom skills in Claude Code?

Your first change

add a hello world function to the main file

Claude will find the file, show you the change, ask for approval (depending on your permission mode), and apply it.

Git, conversationally

what files have I changed?
commit my changes with a descriptive message
create a new branch called feature/quickstart
help me resolve merge conflicts

Permission modes — set this before you go further

Press Shift+Tab to cycle modes. This is the most important control in the product.

Mode Behaviour
default (Manual) Asks before each change
acceptEdits Auto-approves file edits, still asks about commands
plan Claude proposes without editing anything
auto Runs a background safety classifier and blocks risky actions; returns to prompting after repeated blocks
dontAsk Fewer prompts
bypassPermissions No prompts at all

Start in default. Move to acceptEdits once you trust it on a given repo. Use plan for anything you want to review before it happens. Covered in depth in Permissions, sandboxing, and safety.

The commands you’ll use every day

From the shell:

Command What
claude Start interactive
claude "task" Start with an initial prompt
claude -p "query" One-off query, print, exit
claude -c Continue the most recent conversation here
claude -r Resume — picker, or by ID/name
claude -n "name" Name the session
claude --model opus Set the model
claude -w feature-x Start in an isolated git worktree

Inside a session:

Command What
/help All commands
/clear Clear conversation history
/context See what’s loaded and what it costs
/init Generate a CLAUDE.md for this project
/memory View and edit memory files
/usage What’s driving your plan limits
/doctor Diagnose configuration problems
/resume Switch conversations
/exit Quit (or Ctrl+D twice)

Type / to see everything. Tab completes. ↑ gives history. Ctrl+R searches command history across every project.

Six habits worth forming immediately

  1. Be specific. “Fix the login bug where users see a blank screen after entering wrong credentials” beats “fix the bug.”
  2. Break big tasks into numbered steps. Claude follows explicit sequences reliably.
  3. Let it explore first. “Analyse the database schema” before “add a migration.”
  4. Use plan mode for anything risky. Read the plan, then approve.
  5. Run /context when things feel off. Usually the answer is visible there.
  6. Start a new session per task. /clear is cheap; a polluted context is not.

Try it

Exercise 1 — Orientation. Point Claude Code at a codebase you know well. Ask it to explain the architecture. Grade the answer. This calibrates how much to trust it on a codebase you don’t know.

Exercise 2 — Plan mode. Shift+Tab to plan mode. Ask for a non-trivial feature. Read the plan. Note what it got wrong before writing any code — this is the mode’s entire value.

Exercise 3 — Full loop. Have Claude write a small feature, write tests for it, run the tests, fix failures, and commit. One task, one session.

Exercise 4 — Pipe something in.

tail -200 app.log | claude -p "summarise any errors and their likely causes"

Exercise 5 — Name and resume. Start a named session (claude -n auth-work), do some work, exit, and resume it (claude -r auth-work).

Checkpoint

Going deeper