Claude Code has 93 slash commands. Most people use about five of them.
That’s not a guess. A post listing just 16 essential commands pulled nearly 2,000 likes on X — because developers are hungry for this information but drowning in the feature list. And when another developer posted “most people using Claude Code are barely scratching the surface,” hundreds agreed.
So when Anthropic quietly shipped /powerup in v2.1.90 on April 1, 2026 — a built-in interactive tutorial that teaches you Claude Code’s most useful features without ever leaving your terminal — it should have been big news.
It wasn’t. Because /buddy, the Tamagotchi-style coding companion, shipped the same day. A virtual pet beats a tutorial in the attention economy every time.
But /powerup is the more useful feature. And almost nobody knows it exists.
What /powerup Actually Does
Type /powerup in Claude Code and you get a menu of interactive lessons. Not documentation links. Not a README. Actual guided walkthroughs with animated demos that run right in your terminal.
Each lesson covers one feature. You pick the topic, follow along with the animated examples, and try things in your own project. Think Duolingo, but for terminal commands — learn by doing, not by reading.
The system covers 18 topics, organized from beginner fundamentals to advanced workflows. And it’s available to everyone — no Pro subscription required, no extra setup.
Here’s why that matters: Claude Code’s learning curve is steep. Not because the tool is bad, but because it can do so many things that new users don’t know where to start. /powerup fixes that by giving you a guided path through the features that deliver the most value.
The 10 Lessons Worth Your Time
You don’t need to run all 18. Here are the ones that’ll change how you work:
1. Talk to Your Codebase
The foundation. Claude Code can read your entire project — every file, every dependency, every config — and answer questions about it. “/powerup” walks you through how to ask Claude about your code structure, find where things are defined, and understand unfamiliar codebases fast.
This is the feature that makes Claude Code different from ChatGPT. You’re not pasting code into a chat window. The AI already sees everything.
2. Steer with Modes
Claude Code has different operating modes — plan mode, code mode, and others. Plan mode is especially useful: it makes Claude think through a problem before writing any code. For anything touching more than two files, planning first prevents the kind of cascading mistakes that cost you an hour to undo.
3. Undo Anything
One of Claude Code’s best safety features, and one of its least known. If Claude makes a change you don’t like, you can roll it back instantly. The /powerup lesson shows you how the undo system works and why you should commit after every meaningful change — creating save points you can return to.
4. Run in the Background
You can launch Claude Code tasks and keep working while they run. This is huge for longer operations — running test suites, doing large refactors, generating documentation. The lesson teaches you how to kick off background tasks and check on them without losing your flow.
5. Teach Claude Your Rules
This is the CLAUDE.md lesson — probably the highest-leverage feature in the entire tool. You create a file called CLAUDE.md in your project root, and Claude reads it every time you start a session. Put your coding standards, architecture decisions, and project-specific rules in there, and Claude follows them automatically.
One popular tip post (1,777 likes) was nothing but CLAUDE.md rules people should add. That’s how much this one feature matters.
6. Extend with Tools (MCP)
Claude Code supports the Model Context Protocol, which means you can connect it to external tools — databases, APIs, browsers, design tools. The /powerup lesson explains how MCP servers work and walks you through connecting your first one.
If you’ve ever wanted Claude to query your database directly or interact with your staging environment, this is how.
7. Automate Your Workflow (Hooks)
Hooks let you trigger actions automatically at specific points in Claude’s workflow. Set up a hook to run your linter after every file edit. Or validate a deployment script before execution. Or format code on save. The lesson covers the hook system and shows you practical setups.
8. Multiply Yourself with Subagents
This is the advanced stuff — spawning multiple Claude instances that work on different parts of your project simultaneously. One agent writes tests while another refactors the code. The /powerup lesson teaches you when subagents help (large projects with independent modules) and when they don’t (tightly coupled code where changes cascade).
9. Code from Anywhere
Claude Code isn’t just a terminal tool anymore. It runs in VS Code, JetBrains IDEs, as a desktop app, and even in a browser. The lesson covers how to set up each environment and when each one makes sense.
10. Dial the Model
Not every task needs the most powerful (and most expensive) model. The lesson teaches you how to switch between Sonnet 4.6 (your workhorse — use it 80% of the time) and Opus 4.6 (the heavy hitter for complex refactors and architecture decisions). Knowing when to switch saves real money on API usage.
Five Commands You’re Probably Missing
Beyond /powerup, here are slash commands that experienced Claude Code users swear by:
/plan — Forces Claude to think before it codes. For anything complex, this prevents the “it changed 12 files and broke 3 of them” problem.
/clear — Resets your conversation while keeping CLAUDE.md context. Use it every time you start a new task. Old conversation history eats tokens and confuses the AI about what you’re currently working on.
/btw — Lets you ask a side question without interrupting Claude’s current task. It’s like tapping a colleague on the shoulder while they’re coding — “hey, quick question” — without making them lose their place.
/cost — Shows you exactly what you’ve spent this session, broken down by model. The v2.1.92 update added per-model cost breakdowns, so you can see whether that Opus call was worth it.
/buddy — Yes, the Tamagotchi pet. It’s a virtual coding companion that lives in your terminal. Not productive, but surprisingly fun during long coding sessions. Your buddy gains experience as you work and unlocks new abilities. It’s the reason nobody noticed /powerup shipped the same day.
What /powerup Can’t Teach You
The tutorials cover mechanics — what buttons to push, what commands to type. They don’t cover the harder skill: knowing when to use each feature.
That comes from practice. And from learning the patterns that work:
- Start every project by asking Claude to summarize the codebase. This builds its mental model and produces better code from the first prompt.
- Commit after every meaningful change. Claude can introduce regressions, and version control is your safety net.
- Be specific. “Fix the auth” is a bad prompt. “The JWT token expiration check in auth/middleware.ts isn’t accounting for clock skew — add a 30-second grace period” is a good one.
- Use Plan mode for anything touching 3+ files. The upfront planning cost saves you from incorrect changes that are harder to undo than to prevent.
Our Claude Code Mastery course covers these workflow patterns in depth — the kind of judgment calls that turn a Claude Code user into a Claude Code power user. And if you’re still deciding whether Claude Code is the right tool for you, our comparison of Claude Code vs the other vibe coding tools breaks down what each one does best.
How to Get Started
Open your terminal and type:
claude
Once Claude Code is running:
/powerup
Pick a lesson. Start with “Talk to Your Codebase” if you’re new, or “Teach Claude Your Rules” if you want the highest immediate payoff.
The whole system takes maybe 30 minutes to work through the essentials. And unlike reading documentation, you’ll actually remember what you learned — because you did it instead of reading about it.
Claude Code has 93 commands. You don’t need all of them. But you probably need more than five.
Sources:
