Shows help and the available commands in Claude Code.
What the AI does
Lists commands so you can see what you can type.
When to use it
When you are new to Claude Code or forgot a command.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Expecting /help to work in the Claude chat app; it is a Claude Code command.
Example
/help
/clear
Also written: /reset, /new
Starts a new conversation with empty context in Claude Code.
What the AI does
Drops the current conversation so the next task starts fresh, while project memory files stay. You can pass a name to label the previous conversation for /resume.
When to use it
When you switch to a new, unrelated task.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Aliases: /reset, /new.
Common mistakes
Using /clear when you only need to free space; /compact keeps the conversation going.
Clearing in the middle of a task and losing the details Claude needed.
Example
/clear
/compact
Frees up context by summarizing the conversation so far.
What the AI does
Replaces the long conversation with a summary; you can add instructions for what the summary should focus on.
When to use it
When a long session is filling the context window but you want to keep going.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Optional focus instructions can follow the command.
Common mistakes
Expecting every small detail to survive the summary.
Waiting until the context is completely full.
Example
/compact focus on the database changes and the open bugs
/context
Shows how much of the context window is in use, as a colored grid.
What the AI does
Breaks down what is filling the context and shows suggestions and warnings when it is getting full.
When to use it
When answers get worse in a long session and you want to see why.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Pass 'all' to expand the breakdown.
Common mistakes
Ignoring the warning and continuing until the context overflows.
Example
/context
/init
Creates a starter CLAUDE.md guide for your project.
What the AI does
Claude looks at the project and writes a CLAUDE.md file with instructions it will read in later sessions.
When to use it
In the first session in a code repository.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Never reviewing the generated CLAUDE.md; it should match how your team really works.
Example
/init
/memory
Edits CLAUDE.md memory files and manages auto memory in Claude Code.
What the AI does
Opens your memory files so you can refine the standing instructions Claude follows, and lets you view or turn auto memory on or off.
When to use it
After /init, or when Claude keeps forgetting a project rule.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Putting secrets such as passwords in memory files.
Writing vague rules that are hard to follow.
Example
/memory
/model
Switches the AI model Claude Code uses.
What the AI does
Changes the model and saves it as your default for new sessions; with no argument it opens a picker.
When to use it
When a task needs a faster model or a stronger one.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Pass a model name, or run it alone to pick from a list.
Common mistakes
Always using the largest model for simple tasks, which is slower.
Example
/model
/config
Also written: /settings
Opens Claude Code's settings.
What the AI does
Lets you change theme, model, output style and other preferences, or set one directly with key=value.
When to use it
When you want to change how Claude Code looks or behaves.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Example: /config theme=dark. Alias: /settings.
Common mistakes
Changing several settings at once and not knowing which one helped.
Example
/config theme=dark
/resume
Returns to an earlier Claude Code conversation.
What the AI does
Opens a picker of previous conversations so you can continue one.
When to use it
When you come back to work you started earlier.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Starting over from scratch when the old conversation still has the context you need.
Example
/resume
/rewind
Also written: /undo, /checkpoint
Rewinds the conversation and/or code to an earlier point.
What the AI does
Rolls code and conversation back to a checkpoint, or summarizes from a selected message.
When to use it
When Claude went in the wrong direction and you want to undo it.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Aliases: /checkpoint, /undo.
Common mistakes
Rewinding without checking which file changes will be undone.
Example
/rewind
/mcp
Manages MCP server connections in Claude Code.
What the AI does
Opens the list of connected MCP servers, handles sign-in, and can reconnect, enable or disable a server.
When to use it
When you set up or fix tools connected through the Model Context Protocol.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Connecting servers you don't trust; they can see data you give them.
Example
/mcp
/permissions
Manages which tools Claude Code may use without asking.
What the AI does
Opens a dialog to view and edit allow, ask and deny rules for tool permissions.
When to use it
In the first session in a project, or when you are asked for approval too often or too rarely.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Allowing everything to avoid prompts, including commands that can delete files.
Example
/permissions
/add-dir
Adds another folder Claude Code may read and edit in this session.
What the AI does
Gives Claude access to files in the extra working directory you name.
When to use it
When a task needs files from a folder outside the project.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Usage: /add-dir followed by a path.
Common mistakes
Adding a folder that contains private files the task doesn't need.
Example
/add-dir ../shared-library
/export
Exports the current conversation as plain text.
What the AI does
Writes the conversation to a file you name, or lets you copy or save it.
When to use it
When you want to keep or share a record of the session.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Optional: a filename.
Common mistakes
Sharing an export that contains secrets or private data.
Example
/export session-notes.txt
/status
Shows Claude Code's status: version, model, account and connectivity.
What the AI does
Opens the settings screen on the Status tab.
When to use it
When something isn't working and you want to check your setup.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Skipping it when reporting a problem.
Example
/status
/doctor
Also written: /checkup
Runs a setup checkup for Claude Code.
What the AI does
Diagnoses installation, settings, extension and CLAUDE.md problems and proposes fixes that are applied after you confirm.
When to use it
When Claude Code behaves strangely or after installing it.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Alias: /checkup.
Common mistakes
Accepting fixes without reading what they change.
Example
/doctor
/usage
Also written: /cost, /stats
Shows your Claude Code usage.
What the AI does
Shows usage information, including what counts against plan limits on some plans.
When to use it
When you want to know how much you have used.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Aliases: /cost, /stats.
Common mistakes
Assuming every plan shows the same breakdown.
Example
/usage
/code-review
Also written: /review
Reviews the current code changes for correctness bugs.
What the AI does
Checks the diff (or a pull request number you pass) and can apply the findings with --fix.
When to use it
Before you ship a change.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Alias: /review. Effort levels such as low or high can follow the command.
Common mistakes
Treating a clean review as proof there are no bugs; still run your tests.
Example
/code-review high
/security-review
Checks the changes on your branch for security problems.
What the AI does
Reviews the diff between your branch and the default branch for risks such as injection, auth issues and data exposure.
When to use it
Before merging changes that touch logins, data or user input.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Needs a git remote named origin.
Common mistakes
Relying on it as a full security audit.
Example
/security-review
/diff
Shows the changes in your working tree, including Claude's edits.
What the AI does
Lets you review what changed before you keep it.
When to use it
After Claude edits files, before committing.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Committing without looking at the diff.
Example
/diff
/plan
Switches Claude Code into plan mode.
What the AI does
Claude plans the change before editing; you can start straight away with a task, for example /plan fix the auth bug.
When to use it
Before a large or risky change.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Skipping planning on big changes and getting a messy result.
Example
/plan fix the login bug and add a test for it
/plugin
Manages Claude Code plugins.
What the AI does
Opens the plugin menu, or installs, enables or disables a plugin with a subcommand.
When to use it
When you want to add or turn off extra features.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt. Subcommands include list, install, enable and disable.
Common mistakes
Installing plugins from sources you don't trust.
Example
/plugin list
/btw
Asks a side question without adding it to the conversation.
What the AI does
Answers a quick question about the current session while keeping the main conversation unchanged.
When to use it
When you want a quick answer without distracting the main task.
Where it works
Claude Code (Anthropic's coding tool in the terminal): type it at the start of the prompt.
Common mistakes
Using it for instructions that should change the main task.
Every guide and dictionary card is also in the AI Compass app, with saved items, search and “Try it” buttons that open chat, image or video with the example filled in. Reading is free; creating uses credits, and the price is shown before you start.