Most people never make it past level 3. This guide is the ladder. Find your level below, then do the one setup step that moves you up. Every starter file you need is on this page.
01
Prompting
You type, Claude answers. You’re here if Claude feels like a smarter Google.
02
Context
A CLAUDE.md file teaches it your project, rules and preferences once. You’re here if you never re-explain yourself.
03
Tools
Claude Code reads files, writes files, runs commands and searches the web by itself. You’re here if Claude does instead of suggests, but only on your computer.
04
MCP
MCP gives Claude keys to your other apps, like Gmail, Notion or your store.
05
Skills
Repeatable playbooks saved once and used on demand. Your weekly report becomes one request.
06
Subagents
Specialized helpers handling pieces of the work, each with a clean slate. You delegate instead of chat.
07
Agent teams
Several Claude sessions coordinating on shared work. You manage an AI team. Almost nobody is here yet.
How To Use This Guide
Don’t read it all. Find your level, jump to that section and do one setup step. Each level takes minutes to days. One level a week and you’re ahead of almost everyone.
A CLAUDE.md is a plain text file Claude Code reads at the start of every session in that folder. It’s the best 10 minutes in this guide.
Copy this, save it as CLAUDE.md in the folder where you run Claude Code and fill in every bracket. Aim for honest, not perfect.
The Habit That Compounds
Anthropic’s own tip: after every correction, end with "Update your CLAUDE.md so you don’t make that mistake again." Claude adds the lesson to the file, and after a couple of weeks it stops making your specific mistakes.
Want to go deeper on this file? Here’s the most important file in Claude Code.
Claude Code is Claude with hands. It comes with built-in tools to read your files, write new ones, run commands and search the web.
01
Install it
Use the desktop app if you don’t want the terminal, or paste the one-line installer from the setup page. You need a paid plan (Pro, Max, Ultra, Team or Enterprise) or an API account. The Free plan doesn’t include Claude Code.
02
Open it in a folder that matters
Your project, content or business docs. That folder is its workspace.
03
Give it a real job
Phrase it as an order, not a question. "Organize this folder by project and month, and show me what you moved." Watch it work. That moment is the whole level.
Questions Get Suggestions
"Can you suggest a cleanup?" makes Claude talk. "Clean this up and show me what changed" makes Claude work. If it keeps describing instead of doing, you asked a question when you meant to give an order.
MCP servers plug your real apps into Claude. The easiest first ones are the official connectors like Gmail, Google Calendar and Google Drive. Turn them on at claude.ai under Customize (called Plugins on newer accounts), then Connectors. Any connector you add there shows up in Claude Code too, as long as you’re signed in with the same account.
For anything else, add it from the terminal. Here’s Notion as an example, then type /mcp inside Claude Code to sign in and check it’s connected.
claude mcp add --transport http notion https://mcp.notion.com/mcp
Not sure what to add? My 24 Things To Install in Claude guide ranks the MCP servers worth it. Start with one, the tool you touch most every day.
Then test it with a question only your data can answer. "Find every subscription charge in my Gmail from the last 90 days and list them with prices." When the answer comes back from your real inbox, you’ve left level 3.
A skill is a folder with a SKILL.md file inside. Claude reads its description and loads it when the task matches. Anthropic’s rule of thumb: if you do something more than once a day, turn it into a skill.
Save this as .claude/skills/weekly-report/SKILL.md inside your project. The part between the --- lines has to stay at the very top. Fill in the brackets, then ask Claude Code for your weekly report or type /weekly-report.
Copy the pattern for anything you repeat, like inbox triage or a proposal builder.
Subagents are helpers with their own instructions, their own tool permissions and their own clean context. The easiest first one is a read-only reviewer that checks Claude’s work without being able to change anything.
Save this as .claude/agents/reviewer.md. After any big task, say "Have the reviewer check this before we call it done." Or just ask Claude to write a subagent for you.
The Two-Word Power Move
Add "use subagents" to any big request. Anthropic’s power-user tips say it makes Claude throw more compute at the problem by splitting the work across helpers.
More on this in my subagents guide.
Agent teams are several full Claude Code sessions on the same project. One acts as team lead, the teammates share a task list, claim work and message each other directly. It’s still experimental and off by default, and it uses a lot more of your usage than one session.
To try it, add CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS set to 1 in your Claude Code settings, then ask for a team in plain words, like "Spawn three teammates to review this from different angles." Start with research and review jobs, and give each teammate its own files so nobody overwrites anyone.
Want the lighter version first? Run parallel sessions yourself, each in its own worktree with claude --worktree, and keep one session just for reviewing the others. My parallel agents guide walks through it.
You don’t need this level for most work, and what matters here is what you built climbing up.