Claude Code Command Reference

Every slash command, CLI flag, MCP command, keyboard shortcut, and configuration option — in one place.

/slash commands –flags mcp shortcuts CLAUDE.md models

Slash Commands — Complete Reference

/add-dir <path>

Add a new working directory to the current session

/agents

Manage agent configurations

/btw <question> NEW

Ask Claude a quick side question mid-task — no context reset, no re-explaining

/chrome

Configure Claude in Chrome settings

/clear

Clear conversation history and free up context. Aliases: /reset, /new

/color [color|default]

Set prompt bar color for this session: red, blue, green, yellow, purple, orange, pink, cyan

/compact [instructions]

Compress conversation history with optional focus instructions to keep

/config

Open Settings — theme, model, output style. Alias: /settings

/context

Visualize context usage as a colored grid — shows optimization suggestions and capacity warnings

/copy [N]

Copy last response to clipboard. Pass N to copy Nth-latest. Shows picker for code blocks

/cost

Show token usage statistics for the current session

/desktop

Continue session in Claude Code Desktop app. macOS & Windows only. Alias: /app

/diff

Interactive diff viewer — uncommitted changes and per-turn diffs. Arrow keys to navigate

/doctor

Diagnose Claude Code installation — checks config, auth, and updater

/effort [level]

Set model effort: low | medium | high | max | auto. Max requires Opus. No arg = show current

/exit

Exit Claude Code. Alias: /quit

/export [filename]

Export conversation as plain text. With filename writes to file; without opens save dialog

/extra-usage

Configure extra usage to keep working when rate limits are hit

/fast [on|off]

Toggle fast mode — same Opus model, faster output streaming

/feedback [report]

Submit feedback or bug reports about Claude Code. Alias: /bug

/branch [name]

Create a conversation branch at this point. Alias: /fork

/help

Show all available commands and current session info

/hooks

View hook configurations for tool events

/ide

Manage IDE integrations and show connection status

/init

Generate a CLAUDE.md for the current project. Set CLAUDE_CODE_NEW_INIT=true for interactive flow

/insights

Generate report analyzing your Claude Code sessions — patterns, friction points, project areas

/install-github-app

Set up Claude GitHub Actions app for a repository

/install-slack-app

Install the Claude Slack app via browser OAuth flow

/keybindings

Open or create your keybindings configuration file

/login

Switch or add a Claude account

/logout

Sign out of the current Claude account

/mcp

View MCP server status, manage connections and OAuth authentication

/memory

Edit CLAUDE.md files, enable/disable auto-memory, view auto-memory entries

/mobile

Show QR code to download Claude mobile app. Aliases: /ios, /android

/model [model]

View or switch AI model mid-session. Arrow keys adjust effort level

/passes

Share a free week of Claude Code with friends (eligible accounts only)

/permissions

View or update tool permissions. Alias: /allowed-tools

/plan

Enter plan mode — Claude reads and plans but takes no actions

/plugin

Manage Claude Code plugins

/pr-comments [PR]

Fetch GitHub PR comments. Auto-detects current branch PR or pass a URL/number

/privacy-settings

View and update privacy settings (Pro and Max plans only)

/release-notes

View full changelog — most recent version shown closest to prompt

/reload-plugins

Reload all active plugins to apply changes without restarting

/remote-control

Make this session available for remote control from claude.ai. Alias: /rc

/remote-env

Configure the default remote environment for web sessions started with –remote

/rename [name]

Rename the current session. Without a name, auto-generates one from conversation history

/resume [session]

Resume a conversation by ID or name, or open session picker. Alias: /continue

/rewind

Rewind conversation and/or code to a previous point. Alias: /checkpoint

/sandbox

Toggle sandbox mode (supported platforms only)

/schedule [description]

Create, update, list, or run Cloud scheduled tasks conversationally

/security-review

Analyze pending branch changes for security vulnerabilities — injection, auth, data exposure

/skills

List all available skills installed in the current session

/stats

Visualize daily usage, session history, streaks, and model preferences

/status

Open Settings (Status tab) — version, model, account, and connectivity

/statusline

Configure Claude Code’s status line. Describe what you want or auto-configure from shell prompt

/stickers

Order Claude Code stickers

/tasks

List and manage background tasks running in the current session

/terminal-setup

Configure terminal keybindings for Shift+Enter and other shortcuts

/theme

Change color theme — light, dark, colorblind (daltonized), and ANSI variants

/upgrade

Open the upgrade page to switch to a higher plan tier

/usage

Show plan usage limits and rate limit status

/vim

Toggle between Vim and Normal editing modes

/voice

Toggle push-to-talk voice dictation (requires Claude.ai account)

/mcp__<server>__<prompt>

Run MCP server-exposed prompts — dynamically discovered from connected servers

Keyboard Shortcuts

Ctrl+C

Interrupt current operation or clear input

Ctrl+D

Exit Claude Code

Escape

Cancel or clear current input line

↑ / ↓

Navigate through previous commands

Shift+Enter

Insert newline without submitting (after /terminal-setup)

Tab

Autocomplete file paths and slash commands

Slash Commands — Built-in Skills

/loop [interval] [cmd]

Re-run a command on a repeating interval. Default: 10m. Example: /loop 4h /rank-watch

/schedule

Create persistent background triggers on a cron schedule — runs whether or not a session is open

/commit

Stage and create a git commit with an AI-written message following repo conventions

/update-config

Configure settings.json — permissions, hooks, env vars, automated behaviors

/keybindings-help

Customize keyboard shortcuts and chord bindings in ~/.claude/keybindings.json

/seo-audit

Full SEO audit — technical issues, on-page, meta tags, and recommendations

/cold-email

Write B2B cold email sequences with subject lines, hooks, and follow-ups

/social-content

Create and optimize content for LinkedIn, X, Instagram, TikTok

/paid-ads

Google Ads, Meta, LinkedIn ad copy, targeting strategy, and optimization

/ab-test-setup

Plan and design A/B tests with hypothesis, variants, and success metrics

/analytics-tracking

Set up GA4, GTM, event tracking, UTM parameters, and tracking plans

/schema-markup

Add or fix JSON-LD structured data — FAQ, Product, Review, Breadcrumb

/email-sequence

Write nurture sequences, drip campaigns, and triggered email flows

/referral-program

Design referral/affiliate programs with incentive structure and copy

/onboarding-cro

Optimize post-signup onboarding, activation flows, and time-to-value

/find-skills

Discover and install new skills — search available skill library

Permission Modes

default

Standard mode — prompts for approval on sensitive actions

acceptEdits

Auto-approve all file edits, still prompts for Bash commands

dontAsk

Skip confirmation prompts — use in trusted automated workflows

plan

Read-only — Claude plans and explains but takes no actions

auto

Fully autonomous — handles all decisions without prompting

bypassPermissions

⚠ Dangerous — bypasses all checks. Sandboxes only

CLI Flags — Core

-p, –print

Non-interactive mode — prints response and exits. Great for scripts and pipes

-c, –continue

Continue the most recent conversation in the current directory

-r, –resume [id]

Resume a specific conversation by session ID, or open interactive picker

–model <model>

Set model for session. Use alias (sonnet, opus) or full ID

–output-format <fmt>

Output format with –print: text (default) | json | stream-json

–verbose

Show verbose/debug output — useful for diagnosing issues

-n, –name <name>

Set a display name for this session (shown in /resume and terminal title)

–add-dir <path>

Grant tool access to additional directories outside the working directory

–system-prompt <p>

Override the default system prompt for this session

–append-system-prompt

Append text to the default system prompt without replacing it

CLI Flags — Advanced

–permission-mode

Set permission mode: default | acceptEdits | dontAsk | plan | auto

–allowedTools

Whitelist specific tools e.g. “Bash(git:*) Edit Read”

–disallowedTools

Blacklist specific tools from being used in this session

–max-budget-usd

Cap API spend for a –print session (e.g. –max-budget-usd 0.50)

–effort <level>

Set reasoning effort: low | medium | high | max

–mcp-config <file>

Load MCP servers from a JSON file for this session only

-w, –worktree

Create a new git worktree for isolated session work

–fork-session

Create a new session ID when resuming (branches the conversation)

–bare

Minimal mode — skips hooks, CLAUDE.md, memory, LSP. For sandboxes

–dangerously-skip-permissions

⚠ Bypass all permission checks. Sandboxes only

MCP Commands — claude mcp …

mcp list

List all configured MCP servers and their status

mcp get <name>

Show full config and connection details for a specific server

mcp remove <name>

Remove an MCP server by name. Add –scope project for project-level removal

mcp reset-project-choices

Reset all approved/rejected server decisions for the current project

mcp add <name> <cmd>

Add stdio MCP server. Example: claude mcp add gsc — npx @gsc/mcp

mcp add –transport sse

Add SSE MCP server: claude mcp add –transport sse name https://url

mcp add –transport http

Add HTTP MCP server: claude mcp add –transport http name https://url

mcp add -e KEY=val

Pass environment variables to the MCP server process

mcp add-json <name> <json>

Add server using a raw JSON config string

mcp add-from-claude-desktop

Import all MCP servers from Claude Desktop config (Mac / WSL)

mcp serve

Start Claude Code itself as an MCP server for other tools to connect to

MCP Server Scope

–scope user

Available across all projects (default). Stored in ~/.claude/

–scope project

Available to all users of the project. Stored in .mcp.json

–scope local

Available to you only in this project. Stored in .claude/settings.local.json

.mcp.json

Project-level MCP config file. Commit to share servers with your team

CLAUDE.md — Configuration File

./CLAUDE.md

Project-level config — read automatically every session in this directory

~/.claude/CLAUDE.md

Global config — applies to every Claude Code session across all projects

../CLAUDE.md

Parent directory configs are also loaded — chain configs up the directory tree

#import path/to/file.md

Import another markdown file into CLAUDE.md — useful for shared team configs

bash block at top

Bash commands in a fenced block auto-run at session start (e.g. load env vars)

/init

Generate a CLAUDE.md for the current project — Claude reads the codebase first

/memory

Open all CLAUDE.md files in your editor for direct editing

Auto-memory

Claude auto-saves persistent notes to ~/.claude/projects/[path]/memory/

Settings Files

~/.claude/settings.json

Global user settings — applies to all projects

.claude/settings.json

Project settings — commit to share with team

.claude/settings.local.json

Local overrides — add to .gitignore, never commit

model

Default model for sessions

permissions

Allowed/denied tools (persistently)

env

Environment variables loaded at session start

hooks

Shell commands triggered by session events

Hooks

PreToolUse

Runs before any tool call. Can inspect or block the action

PostToolUse

Runs after a tool call completes. Good for logging or notifications

Stop

Runs when Claude finishes responding — trigger alerts or next steps

Notification

Runs when Claude sends a notification event

SubagentStop

Runs when a subagent (spawned by Agent tool) stops

Models Reference

claude-opus-4-6

Most capable — complex reasoning, multi-step tasks. Alias: opus

claude-sonnet-4-6

Balanced speed + capability — default model. Alias: sonnet

claude-haiku-4-5-20251001

Fastest, most cost-efficient — simple tasks and high-volume scripts. Alias: haiku

–model sonnet

Use the alias to always get the latest version of that tier automatically

Non-Interactive Mode — Useful Patterns

claude -p “Summarize this file” < report.txt

Pipe file content to Claude and print the response

claude -p “Fix the bug” –output-format json

Get a structured JSON response — useful for parsing in scripts

claude -p “…” –model haiku –max-budget-usd 0.10

Run a task with a specific model and hard cost cap

claude -p “…” –allowedTools “Read,Glob,Grep”

Restrict to read-only tools — safe for content analysis tasks

cat data.csv | claude -p “Analyze this data”

Pipe any data format — CSV, JSON, logs — directly to Claude

Get the PDF