I’ve been walking a lot of people through setting up Claude Code lately. Many of them aren’t developers. They’re marketers, business people, content creators.
Every time I do this, I find myself repeating the same advice. Open VS Code, open the terminal, type claude, here’s how to skip the permission prompts, here’s how to ask it to do things, here’s what git is and why you don’t need to memorize it.
So I made a cheat sheet.
It covers everything I typically walk people through:
- Installing everything — Homebrew, VS Code, Node, Docker, Claude Code. Platform-specific for Mac, Windows, and Linux.
- Skipping the permission prompts — Auto mode, the
--dangerously-skip-permissionsflag, and how to aliasccso you never type either one again. - Skills to run first — Links to the skills I share with everyone: creating an identity and soul for your agent, setting up persistent memory with AutoMem, building a voice profile, and more.
- How the Human → AI → Human loop works — And why feeding it more context always gets better results.
- Project folder structure — What CLAUDE.md, IDENTITY.md, SOUL.md, and skills/ are and where they go.
- Things to say — Copy-paste prompts for creating content, editing, researching, and the “interview me” prompt that changes everything.
- Git for humans — Vocabulary and plain English prompts. No commands to memorize.
- Memory — How to use AutoMem to start sessions with context and end them by saving what you learned.
- When things go wrong — Quick fixes for every common problem.
I’ll keep updating this as my advice evolves and as Claude Code itself changes. If you’re coaching someone through their first session, send them this link.
The whole thing is one page, collapsible sections, platform switcher at the top. Print it, bookmark it, keep it open in a tab.