Heads up: some links below are referral links. They cost you nothing extra and never change what we recommend. How this site makes money.
The short answer
Slash commands control Claude Code from inside a session — switching models, managing context,
running workflows. Type / to see what's available, or / plus letters to filter.
A command is only recognised at the start of a message; anything after the command name becomes its arguments.
| Command | What it does | When to use it |
|---|---|---|
/init | Generates a starter CLAUDE.md | First thing in a new project |
/memory | Edit persistent project memory | Refining what Claude remembers |
/plan | Switch to plan mode | Before any large change |
/model | Change the active model | Dropping to Sonnet to save usage |
/effort | Adjust reasoning effort | Hard problem, or saving allowance |
/code-review | Multi-agent review of your diff | Before opening a PR |
/security-review | Check the diff for vulnerabilities | Anything touching auth or input |
/rewind | Roll code and conversation back | When it's gone off the rails |
/doctor | Diagnose install and config issues | When something's broken |
Start every project with /init
/init generates a starter CLAUDE.md — the file Claude reads to understand
your project's conventions, structure and constraints. Then use /memory to refine it.
This is the highest-value five minutes you'll spend. A good CLAUDE.md means you stop re-explaining your architecture in every session, and Claude stops making the same wrong assumptions.
Use /plan before anything large
Plan mode makes Claude propose an approach before touching your code.
The value is the same as asking for an outline before a long document: spotting the wrong direction in a ten-line plan takes seconds, spotting it in a finished diff takes a review and a revert.
For any change touching more than a couple of files, /plan first.
/model and /effort control your usage
These two are how you stop burning through your allowance.
/model switches the active model. Opus consumes usage far faster than Sonnet, so dropping to
Sonnet for routine work makes a real difference to how long a session lasts. /effort adjusts
how much reasoning is applied — dial it up for a genuinely hard problem, down for mechanical work.
People who complain about hitting limits are often running Opus at full effort on tasks that didn't need either. See Opus vs Sonnet.
/code-review and /security-review
/code-review runs a multi-agent review of your changes; /security-review checks
the diff specifically for security problems.
Run the security one on anything touching authentication, user input, file handling or permissions. It takes seconds and catches the class of mistake that's expensive to find later.
/rewind is the one people forget
When a session has gone wrong — wrong approach, half-applied changes, a confused conversation
— /rewind rolls both the code and the conversation back to a checkpoint.
Far better than trying to talk Claude out of a hole it's dug. If you're two exchanges into arguing with it, rewind and re-brief instead.
And if the tool itself is misbehaving, /doctor diagnoses installation and configuration
problems before you start debugging your own code unnecessarily.
Common questions
How do I see all Claude Code commands?
Can I create my own commands?
Do commands work mid-message?
Which command saves the most usage?
Claude Code comes with Pro
Included with any paid Claude plan — nothing extra to buy.