Slash Commands

Built-in commands you type at the prompt to control Claude Code’s behavior, manage sessions, and configure settings.

Slash commands are special instructions you type directly at the Claude Code prompt, starting with /. They give you direct control over the session, settings, tools, and integrations — without consuming context or triggering a model call.

> /model
> /compact summarize briefly
> /mcp enable my-server

As you type, Claude Code shows a typeahead menu of matching commands with their descriptions. You can also pass arguments after the command name.

Command types

Internally, commands fall into three categories:

Type	Description
local	Runs synchronously in the CLI process and returns a result
local-jsx	Opens an interactive panel rendered in the terminal
prompt	A skill — sends a prompt to the model

Most built-in commands are local or local-jsx. Skills you create or install are prompt commands.

Session management

/clear — clear conversation history

Clears the conversation history and frees up context. The session starts fresh. Aliases: /reset, /new.

/clear

/compact — compress conversation history

Clears conversation history but keeps a summary in context. Useful when the context window is getting full but you want Claude to retain the gist of what you’ve discussed.

/compact
/compact focus on the authentication changes

Pass optional summarization instructions as an argument.

/resume — resume a previous session

Opens an interactive picker to resume a previous conversation. You can also pass a conversation ID or search term directly.

/resume
/resume auth-refactor

Alias: /continue

/branch — branch the conversation

Creates a branch of the current conversation at this point. Useful when you want to explore an alternative approach without losing the current thread.

/branch
/branch try-a-different-approach

/rewind — rewind to a previous turn

Rewinds the conversation to a previous turn, undoing recent messages and any file changes associated with them.

Configuration

/config — open configuration panel

Opens the interactive configuration panel where you can view and change settings. Alias: /settings.

/config

/model — set the AI model

Sets the model Claude Code uses. Displays the current model and an interactive picker if no argument is given.

/model
/model claude-opus-4-5

/effort — set effort level

Controls how much reasoning effort the model applies. Higher effort means more extended thinking but slower responses.

/effort
/effort low
/effort high
/effort max

Valid values: low, medium, high, max, auto.

/theme — change the theme

Opens the theme picker to change the terminal color scheme.

/theme

/vim — toggle Vim mode

Toggles between Vim and Normal editing modes for the prompt input.

/vim

Tools and integrations

/mcp — manage MCP servers

Opens the MCP server management panel. Use subcommands to enable or disable individual servers.

/mcp
/mcp enable my-server
/mcp disable my-server

/hooks — view hook configurations

Opens a panel showing configured hooks for tool events (PreToolUse, PostToolUse, etc.).

/hooks

/permissions — manage tool permission rules

Opens the permission management panel to configure which tools are always allowed, always denied, or require approval.

/permissions

Alias: /allowed-tools. See Permissions for configuration details.

/ide — IDE integrations

Shows the status of IDE integrations (VS Code, JetBrains) and configuration options.

/ide

Information and diagnostics

/cost — show session cost

Displays the total cost and duration of the current session, including token usage breakdown.

/cost

/status — show session status

Shows the current session status, including model, context usage, and active configuration.

/status

/doctor — diagnose installation

Runs a diagnostic check of your Claude Code installation and settings — verifies authentication, checks for configuration issues, and reports any problems.

/doctor

/help — show available commands

Shows help and lists all available commands with their descriptions.

/help

Code workflows

/review — code review

Runs a code review. If no PR number is given, shows open pull requests to choose from. Uses gh to fetch the PR diff and produces a structured review.

/review
/review 42

/diff — view uncommitted changes

Opens a panel showing uncommitted changes and per-turn diffs for the current session.

/diff

/plan — enable plan mode

Enables plan mode or opens a panel to view the current session plan. In plan mode, Claude proposes steps before executing them.

/plan
/plan implement a dark mode toggle

Content and context

/memory — edit memory files

Opens an interactive panel to view and edit Claude’s memory files (CLAUDE.md, CLAUDE.local.md).

/memory

/skills — list available skills

Lists all skills available in the current session — bundled skills, project skills, and user skills.

/skills

/agents — manage agent configurations

Opens the agent configuration panel. Shows available built-in and custom agents, and lets you manage agent definitions.

/agents

Using commands with arguments

Some commands accept arguments passed directly after the command name:

/compact focus on the database schema changes
/model claude-opus-4-5
/effort high
/resume my-feature-branch

Arguments are passed as a single string after the command name. How each command interprets arguments is command-specific — see the individual command descriptions above.

Typeahead

As you type / at the prompt, Claude Code shows a typeahead list of matching commands. The list updates as you continue typing and includes the command’s description. Press Enter or Tab to complete a command from the list.

For a complete list of all available commands including aliases and argument hints, see the Command reference.