How do I see every Claude Code command?

Type / at the prompt and the menu opens with all of them. Keep typing to filter. /help also lists them with what each one does, and the full list is on this page .

How many commands does Claude Code have?

Claude Code has 118, counting the bundled skills. That count moves as versions ship, which is why this page is dated and the table is read from the reference rather than typed out by hand.

What is the difference between a command and a skill?

At the prompt there is no difference, because you type both the same way. Underneath, core commands are built into Claude Code and bundled skills are loaded. The table on this page marks which is which.

Which command should I learn first?

Learn /context first. It is the one that tells you why the session stopped working, and almost nobody runs it. It is also in the 17 most people never find .

What is the difference between /clear and /compact?

/clear starts a new conversation with an empty context window. /compact keeps the conversation and summarises it to free space. Use /clear when you move to a different task. Use /compact when you are still on the same one and it has just gone long.

Can I make my own commands?

You can. Skills and plugins add their own, and they appear in the same menu. That is also how a command becomes part of your AI OS rather than something you retype.

Do commands work everywhere Claude Code runs?

Most do, but not all of them. Some are tied to where the session lives. /teleport pulls a cloud session into your terminal, and /remote-control exposes a local one to another device. Neither makes sense in a surface that is already the other kind.

Why does a command I read about not exist?

Commands are added and removed between versions, and most lists of them are never checked again. Run claude --version, then check the current reference. A command from an article 6 months old may simply be gone.

How do I go back after Claude changed the wrong thing?

Use /rewind. It rolls the conversation and the code back to an earlier checkpoint. Its aliases are /checkpoint and /undo, which is what most people search for first and why they never find it. More like it are in hidden commands .

What does /btw actually do?

It answers a side question without adding it to the conversation. So you can ask something small while a long run continues, instead of cancelling the run to ask and losing the work.

Do these commands read my AI OS?

They operate on the same session your AI OS is loaded into, so they do in the sense that matters. /memory edits the files directly, and /context shows you whether they loaded at all, which is the first thing to check when Claude seems to be ignoring them .

Is there a command to check my setup is healthy?

Run /doctor. It runs a setup checkup, reports installation and configuration problems, and can fix some of them. It's worth running once while nothing is wrong, so you know what its output looks like normally.

All 118 Claude Code commands, and when to use each

Claude Code has 118 slash commands and most people use about a dozen. Here is every one of them, plus the 26 that map to how a session actually runs.

by Navid Moazzez·Updated 2 Okt 2026·13 min read·

Claude Code commands are the slash commands you type into the prompt to change what the session is doing. Type / and the menu appears. If you haven't used Claude Code yet, start with what it is.

There are 118 of them. Most people meet about a dozen, usually by accident, and never find the ones that would have saved them the most time.

So this is all of them, grouped by the point in the work where you would actually reach for one.

In this guide, I'll show you:

What a Claude Code command is, and how the menu matches what you type
Why the right command at the right moment costs you less than a better prompt
The 26 that map to a normal session, grouped by when you use them
All 118 in one table you can search
The 6 worth setting up before your next session

And every one of them was read from Anthropic's own command reference on September 30, 2026.

key_takeaways.md

TL;DR

Key takeaways

Claude Code commands are instructions to the tool rather than to the model. Type / and the menu opens.

There are 118: core commands built into Claude Code, plus bundled skills you type the same way.

Most of what people try to fix with a better prompt is fixed by a command. /context shows what is filling the window, and /clear gives it back.

Anthropic groups them by when you use them: your first session, during a task, in parallel, before you ship, between sessions, and when something is wrong.

Set up 6 before your next session, starting with /permissions.

What is a Claude Code command?

A Claude Code command is an instruction to the tool rather than to the model. /clear does not ask Claude to forget: it empties the context window.

You type / and a menu opens. Keep typing and it filters.

Some are core commands built into Claude Code. Others are bundled skills, which behave the same way at the prompt but are loaded rather than built in.

Why Claude Code commands matter

Most of what people try to fix with a better prompt is not a prompting problem.

A session that has gone vague usually has a full context window. A run that keeps stalling on approvals usually has no permission rules. Neither gets better by rewording the request.

For example, an hour into a session the answers get worse. The prompt is not the problem. /context shows you what is filling the window and /clear gives it back, and that takes 2 seconds.

So the commands are the controls. Prompting is what you do once the controls are set.

Your first session in a repo

Run these once and the project is set up for every session after it. This is also where your AI OS starts.

Command	What it does
`/init`	Initialize project with a `CLAUDE.md` guide.
`/memory`	Edit `CLAUDE.md` files, enable or disable auto memory, and view auto memory entries
`/mcp [reconnect <server>\|enable\|disable [<server>\|all]]`	Manage MCP server connections and OAuth authentication.
`/permissions`	Manage allow, ask, and deny rules for tool permissions.

During a task

These are the 6 you will use every day, and 3 of them decide what the session costs you.

Command	What it does
`/plan [description]`	Enter plan mode directly from the prompt.
`/model [model]`	Switch the AI model and save it as your default for new sessions.
`/effort [level\|auto\|status\|ultracode [on\|off]]`	Set the effort level: `low` to `xhigh`, `max`, or `auto`; `status` prints it.
`/context [all]`	Visualize current context usage as a colored grid.
`/compact [instructions]`	Free up context by summarizing the conversation so far.
`/btw [question]`	Ask a side question about the current session without adding to the conversation.

Running work in parallel

Use these when 1 task is really several, or when you want your terminal back.

Command	What it does
`/tasks`	View and manage background work in the current session, including subagents that have finished.
`/background [prompt]`	Detach the current session to run as a background agent and free this terminal.
`/batch <instruction>`	Orchestrate large-scale changes across a codebase in parallel.

Before you ship

Read what changed before anyone else has to.

Command	What it does
`/diff`	Review the changes in your working tree, including the edits Claude has made so far.
`/code-review [low\|medium\|high\|xhigh\|max\|ultra] [--fix] [--comment] [pr#\|branch\|path]`	Review the current diff, or a PR number, branch, or path you pass, for correctness bugs.
`/security-review`	Analyze the changes on your current branch for security vulnerabilities.

Between sessions

Work does not end when you close the terminal, and last week's conversation is still there.

Command	What it does
`/clear [name]`	Start a new conversation with empty context.
`/resume [session]`	Resume a conversation by ID or name, or open the session picker.
`/branch [name]`	Create a branch of the current conversation at this point, so you can try a different direction without losing the conversation as it stands.
`/fork [prompt]`	Copy the current conversation into a new background session and keep working here.
`/teleport`	Pull a cloud session into this terminal.
`/remote-control`	Make this session available for Remote Control from claude.ai.