Before you click a single link...
This guide is designed to be read in order the first time. It's not long. Each section sets up the next one. If you skip ahead, you'll end up with a terminal open and no idea why things aren't working - ask me how I know.
What Claude Code Actually Is
Claude Code is not a chatbot. That's the thing no one leads with, and it's the thing that changes everything.
When you chat with Claude at claude.ai, Claude talks to you. That's it. Words in, words out.
Claude Code is different. It can read files on your computer, write and edit files, run commands, and work inside your actual projects - not a sandbox, not a demo environment. Your real files. Your real folders. Your real stuff.
Think of it this way: regular Claude is a brilliant advisor you can talk to through a window. Claude Code is that same advisor, but now they're in the room with you, and they can actually touch things.
Why this matters if you're not a coder
You don't need to know how to code to benefit from Claude Code. You need to know what you want done. "Take everything in this folder and rename the files with today's date." "Read this transcript and pull out every action item." "Build me a one-page website for this offer." That's the level of instruction that works.
The Cost Reality Check
Nobody tells you this upfront and it's annoying. So I will.
Claude Code requires a paid subscription. There is no free access.
Claude Pro - $20/month. This is the entry point. Gives you access to Claude Code with reasonable usage limits for personal projects and exploration.
Claude Max - $100/month (or $200/month for higher limits). If you're using it heavily - multiple projects, long sessions, running it most of the day - you'll eventually hit the Pro ceiling and want to upgrade.
The good news: unlike tools that charge per API call (which can get unpredictable fast), Pro and Max are flat monthly rates. You know what you're paying.
The honest note: long Claude Code sessions use a lot of context. If you work in very long conversations without clearing them, you'll hit usage limits faster. I'll tell you how to manage that in the Slash Commands section.
What You Can Actually Do With This
This is the section I most wanted when I was starting out. Every tutorial I found was written for developers. So here's what Claude Code looks like in the hands of someone who isn't one...
Organize years of files
Gave it the massive data zips from LinkedIn, Facebook, and ChatGPT and had it sort all my writing into Obsidian - tagged and searchable. What would've taken me days took an afternoon.
Process your CRM from DMs
Had it read through my LinkedIn DMs and append notes to the right people files in my CRM. It matched names, pulled context, and updated dozens of records without me touching a single one.
Build web pages with zero code
Sales pages. Resource pages. Landing pages. Complete builds from a description. You don't touch a line of HTML unless you want to.
Create slide decks
From a rough idea to a structured, polished deck. It thinks through flow, suggests structure, and writes slide content. I've used it for client presentations and workshop materials.
Organize Google Drive
Messy folders, inconsistently named files, things saved in the wrong place. Claude Code can audit and reorganize file structures at a scale that would make you want to cry doing it manually.
Extract from transcripts
Got Notion pages full of meeting transcripts? Tell it what you're looking for - action items, a specific person's comments, key decisions - and it pulls exactly that.
The honest scope check
Claude Code is not magic. It hallucinates sometimes. It makes mistakes. It occasionally does something you didn't ask for. The more specific your instructions, the better it does. Think of it like delegating to a very capable person who needs clear direction and benefits from you checking their work.
Install It (Pick Your Path)
Start by picking your computer, then choose your path. Three options - pick whatever feels least terrifying.
Claude Desktop App
Download the app, click the Code tab. No terminal. No installs. Start here if you're not sure.
Terminal
One command to install, then type claude to start. More control. Worth it once you do it once.
VS Code
A free visual editor with Claude Code built in. Good if you want to see your files alongside Claude.
Claude Desktop App (Mac, easiest)
Download the Claude Desktop App
Go to claude.ai/download and download the Mac version. Install it like any other app.
Sign in and click the Code tab
Open Claude, sign in with your Pro or Max account, and look for the Code tab at the top. That's Claude Code - full capabilities, visual interface, no terminal required.
Open a folder and start
Point it at a folder on your computer and start telling it what to do. You'll see diffs (what changed) visually before anything is applied.
Terminal (Mac)
Open Terminal
Press Cmd + Space, type Terminal, press Enter. A window opens with a blinking cursor. You're in.
Install Claude Code with one command
Paste this exactly and press Enter:
That's it. No separate software to install first. It handles everything. Wait for it to finish.
Start Claude Code
It'll ask you to log in with your Anthropic account (same one you use for Claude Pro/Max). Once you see >, you're running Claude Code.
Prefer Homebrew?
If you already use Homebrew, you can install with: brew install --cask claude-code - then type claude to start.
VS Code (Mac)
Download VS Code
Go to code.visualstudio.com and download the Mac version. Free. Install it like any app.
Install Claude Code inside VS Code
Press Cmd+Shift+X to open Extensions. Search Claude Code and install the official Anthropic extension.
Open a project and start
File - Open Folder to open whatever you're working on. Press Cmd+Shift+P, type Claude Code, and select Open in New Tab. Sign in when prompted.
Claude Desktop App
Download the app, click the Code tab. No terminal. No installs. Start here if you're not sure.
PowerShell
One command to install, then type claude to start. More control. Worth it once you do it once.
VS Code
A free visual editor with Claude Code built in. Good if you want to see your files alongside Claude.
Claude Desktop App (Windows, easiest)
Download the Claude Desktop App
Go to claude.ai/download and download the Windows version. Install it normally.
Sign in and click the Code tab
Open Claude, sign in with your Pro or Max account, and click the Code tab. Full Claude Code, no terminal needed.
PowerShell (Windows)
Open PowerShell as Administrator
Press the Windows key, type PowerShell, right-click it, and choose "Run as Administrator."
Install Git for Windows first (if needed)
Windows requires Git. Go to git-scm.com/downloads/win and install it if you don't have it. Restart PowerShell after.
Install Claude Code
Wait for it to finish.
Start Claude Code
Log in with your Anthropic account when prompted.
Windows note
If you get permission errors, make sure you right-clicked PowerShell and chose "Run as Administrator." Windows needs explicit permission to install programs globally.
VS Code (Windows)
Download VS Code
Go to code.visualstudio.com, download the Windows version, and install it. Free.
Install the Claude Code extension
Press Ctrl+Shift+X to open Extensions. Search Claude Code and install it.
Open a project and start
File - Open Folder, then Ctrl+Shift+P, type Claude Code, select Open in New Tab.
CLAUDE.md: The Thing Nobody Told Me About
If you only implement one thing from this entire guide, make it this.
CLAUDE.md is a plain text file that tells Claude Code how to work with you. It's read at the start of every session - your preferences, your projects, your context - so you're not re-explaining yourself every single time you open a new conversation.
There are two of them, and they serve different purposes:
The Home-Level CLAUDE.md
Lives at the root of your computer. This is about you - how you think, how you communicate, what you're working on broadly, what Claude should never do without asking. Claude reads this every single session, regardless of what project you're in.
Location on Mac: ~/.claude/CLAUDE.md - on Windows: your home directory
The Project-Level CLAUDE.md
Lives inside a specific project folder. This is about that project - the goal, the files involved, the rules for this specific work. When you're in that folder, Claude reads both files. The project one takes priority if there's ever a conflict.
Here's a starter template. Copy it, fill in your details, save it as CLAUDE.md in your home folder. This is the version I wish someone had handed me on day one.
How to create the file
Easiest way: Copy the template above. Start a Claude Code session. Tell Claude: "Create a CLAUDE.md file in my home directory with this content:" and paste it. Claude Code will create the file for you and put it in the right place.
For a project-level file: Navigate to your project folder and tell Claude: "Create a CLAUDE.md in this folder that describes this project." Give it a few sentences about what you're building. Review what it generates.
Slash Commands - Your Control Panel
Once you're in a Claude Code session, slash commands are how you manage what's happening. These were mentioned everywhere when I was learning but nobody explained them. Here they are, organized by what you actually need them for.
| Command | What it does and when to use it |
|---|---|
| Navigation & Basics | |
| /help | Shows all available commands. Type this when you forget what's available - which will happen often at first, and that's fine.
Use anytime you're lost.
|
| /exit
also: /quit
|
Exit the Claude Code session cleanly.
Use when you're done. Or just close the window.
|
| /resume
also: /continue
|
Pick up where you left off in a previous session. Shows a list of past sessions to choose from.
Use when you want to continue work from an earlier conversation without starting from scratch.
|
| Managing Context (This Is the Important Stuff) | |
| /clear ★
also: /reset, /new
|
Wipes the conversation and starts completely fresh. Claude forgets everything from this session.
Use when a session has gotten long, confused, or sideways. Don't wait until it's fully broken - clear it before the wheels come off. Your files stay as they are; only the conversation is cleared.
|
| /compact | Compresses the conversation to save space without losing the thread entirely.
Use when your session is getting long but you're not done and don't want to lose context. Good for builds that take a while.
|
| /context | Shows you a visual of how full your context window is, with warnings if you're getting close to the limit.
Use when things start feeling sluggish or responses get weird. If it's over 80%, consider /compact or /clear.
|
| /cost | Shows your token usage stats for the current session.
Use when you're curious how much of your plan you've used, or if you want to keep sessions efficient.
|
| /usage | Shows your overall plan usage limits and where you stand on rate limits.
Use when you want to know how close you are to hitting your monthly ceiling.
|
| Review, Undo, and Safety | |
| /plan | Puts Claude into plan-only mode. It tells you what it's going to do before doing any of it.
Use at the start of anything involving lots of file changes. Review the plan, say go ahead, then watch it work. Huge for non-coders who want to stay in control.
|
| /diff | Opens an interactive viewer showing what Claude changed - file by file, line by line.
Use after Claude finishes a task to see exactly what it did before you consider it done.
|
| /rewind ★
also: /checkpoint
|
Rewinds the conversation AND the files back to a previous point. Undoes Claude's changes.
Use when Claude did something you didn't want and you need to go back. This is the safety net. It's real - use it freely.
|
| /security-review | Scans recent changes for security issues - injection risks, exposed data, auth problems.
Use if Claude built something that will be public-facing, like a web page or form.
|
| Setup and Memory | |
| /init | Scans the current project folder and auto-generates a CLAUDE.md for it.
Use at the start of any new project. Review what it generates - it might miss things, but it's a solid first draft.
|
| /memory | Opens your CLAUDE.md files so you can read or edit them from inside a session.
Use when you want to update your preferences or project notes without leaving Claude Code.
|
| /doctor | Runs a health check on your Claude Code installation and flags configuration problems.
Use when something feels off but you can't pinpoint what. Good first step in troubleshooting.
|
| /config
also: /settings
|
Opens the settings interface - theme, model, output style, and other preferences.
Use when you want to change how Claude Code looks or behaves.
|
| Handy Extras | |
| /btw | Ask a quick side question without it becoming part of the conversation. The question and answer disappear after you dismiss them.
Use when you have a quick "wait what was that file called again" moment and don't want to derail what Claude is working on.
|
| /copy | Copies Claude's last response to your clipboard. If there are code blocks, shows a picker so you can grab just one piece.
Use when you want to paste something Claude wrote into another app.
|
| /export | Exports the full conversation as plain text - to clipboard or a file.
Use when you want to save a record of a session, or share what happened with someone.
|
| /model | Switch which Claude model you're using mid-session.
Use when you want to try a faster model for simple tasks or a more powerful one for something complex.
|
| /feedback
also: /bug
|
Report a bug directly to Anthropic.
Use when something feels genuinely broken, not just confusing. Save this for actual malfunctions.
|
The thing that tripped me up
I didn't understand how much /clear matters until my sessions started going sideways. Long sessions accumulate context - Claude starts "remembering" earlier parts of the conversation in ways that muddy what you're asking now. When in doubt, clear it and start the new task fresh. It's not failure. It's hygiene. And /rewind is your undo button - don't be afraid to use it.
When It Breaks and You Have No Idea Why
It will break. Not often, but it will. Here's how to think about it before you start Googling yourself into a spiral.
Claude seems confused or is going in circles
Your session has too much accumulated context. Type /clear, then restate your goal from scratch - clearly and specifically. This solves it 80% of the time.
Claude is doing something you didn't ask for
Stop it immediately - type stop or press Ctrl+C. Then be more specific about what you want. Vague instructions are the number one cause of Claude going off in an unintended direction. Use /rewind to undo what it already did.
Claude is editing the wrong file
This one bites people early. Before any big task, ask: "What file are you about to edit? Show me the full path." Make it say the exact file before it touches anything. Getting specific about paths prevents a lot of heartbreak.
You can't see what changed
Use /diff to see exactly what Claude did - file by file. In the Desktop App you see this visually before applying changes. In the terminal, /diff is your friend.
Nothing is working and you don't know why
Run /doctor first. Then check claudefa.st/blog/guide/troubleshooting - it has a run-these-in-order format that covers the most common "everything is broken" scenarios without requiring you to diagnose the root cause yourself.
The habit that saves the most headaches
Before Claude does anything big - reorganizing a folder, rewriting a document, touching files you care about - make a backup copy first. Ten seconds of copying to a backup folder has saved many people a lot of pain.
One Starter Prompt to Save Right Now
Copy this. Paste it at the start of your first real Claude Code session. It sets context and gets Claude oriented fast.
I'm new to Claude Code. I'm not a developer. Before we start any task, I need you to: 1. Confirm in plain English what you're about to do 2. Tell me exactly which files you'll be touching 3. Wait for my "go ahead" before making any changes My goal today is: [describe what you're trying to accomplish] My computer is a [Mac / Windows]. Ask me if you need any other context.
Adapt it to your situation. The point is making Claude slow down and confirm before it acts - especially while you're still learning what it's capable of.
Resources Worth Your Time
Now that you're set up and oriented - here's what I actually found useful. Curated, not comprehensive.
Non-Coder's Intro to Claude Code
Two-part series. Screenshot-heavy. Shows the exact "before you click go" moment that most guides skip entirely.
Part 1 Part 2Everyone Should Be Using Claude Code
The Lenny's Newsletter take - solid context on why this matters even if you're not building software.
WatchClaude Code Walkthrough
Practical, no-fluff walkthrough. Good for seeing what a real session looks like before you start your own.
WatchA Beginner's Guide to Claude Code (No Coding Required)
Screenshot-heavy, written for non-devs. The visual walkthrough of first-run setup is particularly good.
ReadHow to Use Claude Code for Non-Engineering Teams
Practical framing for people in marketing, ops, and other non-technical roles.
ReadVibe Coding Is the Biggest Unlock for Non-Technical Founders
Jodie Cook on why this moment matters for people who build without engineering backgrounds.
ReadEveryone Should Be Using Claude Code
The written companion to the video. Good if you'd rather read than watch.
Readr/ClaudeAI
Where real users troubleshoot in real time. Search before you post - most questions have been asked.
VisitClaude Code for Non-Coding Work: A Complete Guide
Long thread with practical examples from non-devs using Claude Code for real projects.
Read threadAnthropic Official Discord
Direct access to other Claude Code users and sometimes Anthropic staff. Useful for edge cases.
Joinclaudefa.st - 5 Fixes That Solve 90% of Problems
Run-these-in-order format. No diagnosis required.
Openclaudelog.com - Troubleshooting Guide
More comprehensive. Worth bookmarking for the "agent stuck" section specifically.
OpenAnthropic Official Docs
Technical and authoritative. Use as a last resort when community advice hasn't solved it.
OpenAbout the person who made this
Deb Haas spent 24 years in Global HR at Accenture before getting laid off on March 1, 2024. She has not gone back to a regular job. She has also not been calm about it.
She's 56, has been remote since 2006, lives in Minneapolis with her husband (Smidge) and three cats, and is fully in her feral feminist years. She taught herself AI accidentally while figuring out what to do with her life, built a business through trial and error, and only started charging when her creditors made clear they wouldn't accept vibes as payment.
Her love languages are movies, music, and books. Her newsletter is Mostly Human (named by a LinkedIn friend), her community is Joy Prompt Club (after Joy Luck Club), her other project is AI Confidential (after L.A. Confidential). GenXer. It tracks.
If this guide helped, her free weekly newsletter is about staying human in an AI world.