Cheat Sheet

10 Claude Code Commands You're Not Using

Anthropic have quietly built 75+ commands into Claude Code. You just type a slash. Here are the ten worth knowing — use all of them and you're in the top 1%.

>/
/init
/security-review
/context
/rewind
+ 70 more…

A slash command is a shortcut you fire inside Claude Code — type / in the terminal or the VS Code extension and a menu pops up. Most people use about three of them. These are the ten that actually change how you work.

The Ten

1/initGives Claude a memory of your project.

It reads your whole codebase and writes a claude.md — how the project works, how to run it, the conventions — so every future session is instantly up to speed instead of relearning it.

2/security-reviewChecks your code for security holes before you push it live.

A read-only pass over your changes for the classic vibe-coding mistakes — exposed keys, missing auth, injection — with a fix for each.

How the security review works
3/contextShows you exactly what's eating your tokens.

A visual breakdown of what's filling your context window, so you can see what's quietly making your chat slow and expensive.

Why you hit your limit so fast
4/rewindRewind the codebase to undo a change you don't want.

Jump back to an earlier checkpoint in the session and undo the edits since — your safety net when it goes the wrong way.

5/resumeDrop back into a chat you closed days ago, right where you left off.

Pick a past session by name, and carry on with the full context intact — nothing lost.

6/remote-controlCarry your live session straight to your phone.

Syncs this session to claude.ai so you can keep going from your browser or phone and hand back to your laptop seamlessly. Short alias: /rc.

7/loopRuns a check on repeat while you go and do something else.

Give it a prompt and an interval and it runs on a loop — great for watching a deploy, or re-running a check until it passes.

7 ways to actually use loops
8/compactSummarises everything so far to free up room so you can keep working.

Condenses the conversation — keeping your claude.md, rules and memory — so a long chat doesn't grind to a halt. You don't have to start a new one.

What the context window is
9/deep-researchGoes off, searches the web, and comes back with a full cited report.

Fans out across multiple searches, cross-checks the sources against each other, and synthesises an answer with citations — not one quick lookup.

10/code-review --fixFinds the bugs in your code and fixes them for you.

Reviews your current diff for correctness bugs and cleanups. Add --fix and it applies them; --comment posts them on your PR instead.

How to run them

Just type / in Claude Code — in the terminal or the VS Code extension — and the menu appears. Start typing the name to filter. Type /help to see the lot.

And There Are Loads More

/usagewhat you've spent and how close you are to your limit
/simplifya cleanup-only pass on your changed code (no bug-hunting)
/agentsspin up subagents to keep your main chat sharp
/scheduleset up routines that run on a cron in the cloud
/diffscroll back through every change, turn by turn
/modelswitch between Sonnet and Opus mid-task

Plenty of the 75 you'll never need. These ten are the ones worth committing to memory.