Step 1 Installation

Installing Claude Code

Two options per platform. Pick the one that matches your comfort level. No coding experience required.

Option A: Desktop App Easiest

Download the installer from claude.ai/download. It's a standard Windows .exe file.
Run the installer and follow the on-screen prompts. No configuration needed.
Sign in with your Anthropic account when the app opens.

Done. The desktop app skips the terminal entirely. Use this if you prefer not to use the command line.

Option B: Terminal Install Recommended for full features

1

Install Node.js

Go to nodejs.org/en/download and download the Windows Installer (.msi), LTS version. Run it and click Next through all screens. This installs both Node.js and npm together. Already have Node.js? Run node --version to confirm, then skip to step 2.

2

Open PowerShell

Press Win + X and select Windows PowerShell or Terminal. You should see PS C:\Users\YourName> at the start of each line.

3

Fix the execution policy (do this once, required on most machines)

Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned

Type Y and press Enter when prompted. This only changes the policy for your user account. No admin rights required.

4

Install Claude Code

npm install -g @anthropic-ai/claude-code

5

If you see an "allow-scripts" warning

Run this to approve the post-install setup step:

npm approve-scripts @anthropic-ai/claude-code

6

Launch Claude

Close PowerShell, open a new window, type claude and press Enter.

Seeing "running scripts is disabled on this system"? You skipped step 3. Run the Set-ExecutionPolicy command above first, type Y when prompted, then try the npm install again.

Make sure you're in PowerShell, not CMD. You should see PS C:\Users\YourName> at the start of each line. No "PS"? Close the window and reopen PowerShell.

Got a "claude is not recognized" error? Restart your computer and try again. That fixes it 90% of the time.

Option A: Desktop App Easiest

Download the installer from claude.ai/download. Look for the macOS download and grab the .dmg file.
Open the .dmg, then drag the Claude icon into your Applications folder.
Open Claude from Applications (or use Spotlight: Cmd + Space, type "Claude") and sign in with your Anthropic account.

Done. The desktop app skips the terminal entirely. Use this if you prefer not to use the command line.

Option B: Terminal Install Recommended for full features

1

Open Terminal

Press Cmd + Space, type Terminal, and press Enter. Or go to Applications > Utilities > Terminal. You'll see a prompt that ends in % or $.

2

Install Homebrew (skip if you already have it)

Homebrew is the standard Mac package manager. Paste this into Terminal and press Enter:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

It will ask for your Mac password. Type it and press Enter. Nothing will appear as you type — that's normal. Follow any prompts until it finishes. Already have Homebrew? Run brew --version to confirm, then skip to step 3.

3

Install Node.js

brew install node

This installs both Node.js and npm. Already have Node.js? Run node --version to confirm, then skip to step 4.

4

Install Claude Code

npm install -g @anthropic-ai/claude-code

5

Launch Claude

Type claude and press Enter. It will walk you through signing in on your first run.

Seeing "permission denied" or "EACCES" errors? This usually means npm's global folder needs fixing. Run sudo npm install -g @anthropic-ai/claude-code instead and enter your Mac password when prompted.

Homebrew says "command not found: brew" after installing? Close Terminal completely, reopen it, and try again. On Apple Silicon Macs (M1/M2/M3), Homebrew installs to /opt/homebrew. If it still doesn't work, follow the "Next steps" instructions Homebrew printed at the end of the install.

Got a "claude: command not found" error? Close Terminal and reopen it. That refreshes the PATH so it finds newly installed tools.

Step 1 of 5

Step 2 Terminal Basics

Basic terminal commands

Eight commands. That's all you need. Once you type claude you're back in plain English. The terminal is just the on-ramp. These work the same in PowerShell (Windows) and Terminal (Mac).

pwd

Show your current folder location

ls

List files and folders in the current directory

cd FolderName

Move into a folder. Use cd .. to go back up one level

mkdir FolderName

Create a new folder in your current location

claude

Launch Claude Code in the current folder

Ctrl + C

Stop whatever is currently running immediately

↑ arrow key

Recall your last typed command to reuse it

exit

Close the current Claude session or terminal window

Always navigate to your project folder first. Claude works within the folder you launched it from. Use cd to get there, then run claude.

Once Claude Code is running, you won't need terminal commands. Just type what you want in plain English and press Enter.

Step 2 of 5

Step 3 Prompting

How to prompt Claude effectively

Claude is only as good as the instructions you give it. Almost every wrong output, error, or misunderstanding can be fixed with a better prompt, not a restart.

The prompting formula

Context + Clear action + Expected output

Tell Claude what you're working with, what you want done, and what the result should look like.

Weak vs. strong prompts

Make a spreadsheetToo vague

Create a weekly inventory tracking spreadsheet for pre-rolls with columns for SKU, quantity, batch date, and machine used. Format it so it's ready to fill in.Clear

Fix thisToo vague

I have a Python script that reads from a CSV and calculates daily output totals. It crashes with a KeyError on line 12. Here's the error: [paste error]. Fix it and explain what was wrong.Clear

✓ Do

Specify the format you want: "give me a numbered list", "output as a table"

Paste the actual error message or file content

Say what you'll use the output for: "this goes in an email to my manager"

✗ Avoid

One-word requests with no context

Assuming Claude knows your setup without telling it

Starting over when something's wrong. Just correct it in the same conversation.

Not sure how to do something? Just ask Claude.

Claude can walk you through anything, including its own features

Don't get stuck. If you don't know how to do something (whether it's a terminal command, a file format, or a task at work), describe it to Claude in detail. The more context you give, the better the answer.

Include: what you're trying to do, what you've already tried, what happened, and what you expected. You can paste screenshots, error messages, or file contents directly into the conversation.

📋

Paste error messages

Copy the full error text and paste it in. Claude reads every line.

🖼️

Share screenshots

Drag an image into the chat. Claude can see what's on your screen.

📁

Upload your files

Drop a CSV, script, or document into the conversation for Claude to read.

✏️

Correct and continue

If the output is wrong, say why and ask again. Don't start over.

"I got this error, what does it mean?" "Walk me through this step by step" "Here's a screenshot. What should I do?" "That's not quite right, try again with..."

These prompts work well when you're stuck

Useful prompts to try in Claude Code

Look at the files in this folder and tell me what this project doesOrientation

I got this error when I ran the script. What does it mean and how do I fix it? [paste error]Debugging

Write a script that reads all the .csv files in this folder and outputs a summary reportAutomation

Explain what this script does in plain English, step by stepUnderstand

Step 3 of 5

Step 4 Skills

What are skills & how to install one

Skills are add-ons that give Claude specialized knowledge for your specific workflows. They load automatically every session. Think of them as pre-built expertise for your exact job.

Without a skill

Claude is general-purpose. You re-explain context, formats, and preferences every single conversation.

With a skill

Claude already knows your workflow, naming conventions, and output format. Every time, automatically.

How to install a custom skill (provided by your team)

1

Enable code execution

Go to claude.ai → Settings → Capabilities and turn on Code execution and file creation. Skills won't appear without this.

2

Open your Skills page

Navigate to claude.ai → Customize → Skills. You'll see active skills and a "+" button to add more.

3

Upload the skill ZIP file

Click "+" → Create skill → Upload a skill. Select the .zip file your team provided. It appears in your list.

4

Toggle it on

Use the toggle switch next to the skill name to enable it. Claude will use it automatically whenever it's relevant.

You don't need to "activate" the skill every session. Once toggled on, it's always available. You can also invoke it explicitly: "use the [skill name] skill to..."

Security note: Only install skills provided by your team or Anthropic directly. Skills can run code, so only use ones from trusted sources.

Step 4 of 5

Step 5 Assignment

Your first skills assignment

You've been given a skill ZIP file alongside this guide. Your job is to install it, understand it, and customize it so it works for your specific role and workflow.

What you received

A custom Sorting Robotics skill file

This skill was built to give Claude a way to help you build your own code. Your task is to make it your own. A skill that fits your role will save you hours every week.

Complete all four tasks

Install the skill

Follow the steps from Step 4 to upload and enable the provided skill ZIP. Confirm it appears in your skills list with the toggle on.
Install

Stuck? Try asking Claude:

"I'm trying to install a skill ZIP file in Claude. I'm on Step [X] and I see [describe what's on screen]. What should I do next?"
Understand what it does

Open the SKILL.md file inside the ZIP and read it. Then ask Claude to explain what the skill does and when it would trigger, in plain English.
Understand

Useful prompts for this task:

"Here is the content of my SKILL.md file: [paste contents]. Explain what this skill does, when it activates, and what it changes about how you respond."

"What parts of this skill are most relevant to my day-to-day work as a [your role]?"
Edit it to fit your needs

Open the SKILL.md file in any text editor (Notepad works). Add at least one customization: your machine type, your team's naming convention, a preferred output format, or a task you repeat often. Re-zip and re-upload the skill.
Edit

Ask Claude to help you customize it:

"Here is my current SKILL.md: [paste it]. I work with the Stardust machine and track production by shift. Edit this skill so Claude always uses those conventions when helping me."

"Add a section to my skill that tells Claude to always format reports as a table with columns for date, batch, operator, and output count."
Write your About Me section

Your agent works better when it knows who you are. Ask Claude to interview you and generate an About Me section for your CLAUDE.md. You'll use this in Module 3 when you set up your security config, so save it somewhere you can find it.
Edit

Paste this prompt into Claude:

"I need to write an About Me section for my CLAUDE.md file so my agent understands my context at work. Ask me questions about my role at Sorting Robotics, what systems I use day-to-day, my technical background, and what kinds of tasks I want to automate. Then generate the section in a format I can copy directly into the file."

0 of 4 tasks completed

✓ Module 1 complete

You've installed Claude Code, learned the terminal basics, installed a skill, and written your About Me section. Save that About Me. You'll need it in Module 3. Next up: Prompting & Scripting.

Reminder: If you get stuck at any point, describe exactly what you see on screen and paste any error messages into Claude. Be specific: "it's not working" is hard to debug, but "I see this error after clicking Upload: [error text]" gets you an answer in seconds.

Step 5 of 5