Goal: In about 15 minutes, you will install Claude Code, sign in, and complete your first AI-assisted change. This guide covers only what is needed on day one; links to deeper material appear at the end.
Step 1: Install Claude Code (2 minutes)
Pick the command that matches your machine and paste it into a terminal.
Your setup | Command |
macOS / Linux / WSL | `curl -fsSL https://claude.ai/install.sh |
Windows (PowerShell) | |
Homebrew |
|
npm (Node 18+) |
|
Tip: Do not run the npm install with sudo, as this causes file-permission problems later.
Verify the installation:
claude --version
Step 2: Sign in (2 minutes)
How you authenticate depends on how your organization has provisioned access. Pick one of the following:
If your organization uses… | Do this |
Claude Enterprise seats | Run |
An Anthropic Console API key | Run |
Amazon Bedrock | Run |
Google Vertex AI | Run |
If you are unsure which method applies, check with your administrator or try /login first, which is the default for most organizations.
Step 3: Open a project (1 minute)
cd path/to/your/project claude
Claude Code reads your files on demand, so there is nothing to upload or attach. Simply start it from inside the repository you want to work on.
Step 4: Understand the one safety rule (1 minute)
Claude will always ask before changing a file or running a command. When it proposes a change, you will see a diff and three choices:
Yes — apply this single change.
Yes, and don't ask again for edits — auto-approve file edits for the remainder of the session.
No — reject the change and explain why.
You can press Shift+Tab at any time to cycle between modes (Plan → Accept Edits → default). If a change produces an unexpected result, your git history remains untouched until you commit, so git checkout will undo it.
Step 5: Five things to try first
Copy and paste these into the prompt one at a time. Each example demonstrates a different core capability.
1. Get oriented
Give me a 5-bullet summary of what this codebase does and where the entry point is.
Why this is useful: It shows how Claude explores files on its own, without being pointed at specific paths.
2. Find something
Where is user authentication handled? Show me the file and the key function.
Why this is useful: It is often faster than grep when you do not know the exact symbol or filename.
3. Make a safe edit
Add a docstring to the function <name> in <file>. Keep it to 2 lines.
Why this is useful: It lets you practice reviewing and approving a diff on a low-risk change.
4. Fix something real
This test is failing: <paste the error>. Find the cause and fix it.
Why this is useful: This is the primary workflow—you describe the symptom, and Claude investigates and proposes a fix.
5. Let it handle git
Stage my changes and write a commit message that follows our existing style.
Why this is useful: Claude can run git on your behalf (with approval) and will match your repository's existing commit conventions.
Step 6: Before you finish for the day
Run this once per project:
/init
This generates a CLAUDE.md file at your project root that captures your codebase's conventions. Claude reads it automatically at the start of every session, so future responses will already be tailored to your project. It is the single highest-value setup step you can take.
Next steps
You have now installed Claude Code, authenticated, made an edit, and set up project memory. When you are ready to go further, the official documentation covers memory and CLAUDE.md, slash commands, and interactive-mode shortcuts in depth.
If you ran into problems during install or login, see the troubleshooting guide.