Claude Code — Command Cheat Sheet (v2.x)

Complete command reference for Claude Code v2.x (docs current as of Aug 2026). Covers: slash commands (all), CLI subcommands, key flags, keyboard shortcuts, and input prefixes. Quick-lookup cheat sheet — explanations are condensed; see the official CLI reference for details.

1. Slash Commands — Session & Conversation

Command	What it does	Example
/add-dir <path>	Grant file access to an extra working directory for this session (monorepos)	/add-dir ../shared-lib
/background [prompt]	Detach current session to run as a background agent, freeing the terminal	/background finish the tests
/branch [name]	Branch the conversation at this point to try another direction (keep the original)	/branch try-approach-B
/btw [question]	Ask a side question without adding to the conversation context	/btw what's the convention for error types?
/cd <path>	Move the session to a new working directory, keeping the conversation	/cd src/api
/clear [name]	Start a fresh conversation (empty context); keep project memory	/clear
/compact [focus]	Summarize the conversation to free context (CLAUDE.md survives)	/compact focus on the auth logic
/context [all]	Visualize context usage as a colored grid with optimization tips	/context
/copy [N]	Copy the last (or Nth-latest) assistant response to the clipboard	/copy 2
/export [filename]	Export the conversation as plain text (file or clipboard)	/export session.txt
/focus	Toggle focus view: only last prompt, tool summary, and final response	/focus
/fork [prompt]	Copy this conversation into a new background session and keep working here	/fork run the perf analysis
/goal <condition>	Keep working across turns until the condition is met	/goal all tests pass
/recap	Generate a one-line summary of the current session	/recap
/rename [name]	Rename the current session (shown on prompt bar, in /resume)	/rename auth-refactor
/resume [session]	Resume a conversation by ID or name, or open the picker	/resume auth-refactor
/rewind	Rewind conversation and/or code to a previous checkpoint (aliases: /checkpoint, /undo)	/rewind
/exit	Exit the CLI (in a background session: detaches, session keeps running)	/exit
/teleport	Pull a Claude Code web session into this terminal	/teleport
/remote-control	Make this session controllable from claude.ai / the Claude app	/remote-control

2. Slash Commands — Review & Code Quality

Command	What it does	Example
/code-review [level] [--fix] [pr#\|branch\|path]	Review diff/PR/branch for correctness bugs + cleanups (alias: /review)	/code-review high 1234 --fix
/security-review	Analyze branch diff for security vulnerabilities (injection, auth, data exposure)	/security-review
/simplify [target]	Run 4 parallel review agents over changed code and apply cleanup fixes	/simplify src/utils.ts
/ultrareview [PR\|branch]	Deep multi-agent review in a cloud sandbox	/ultrareview main
/diff	Interactive diff viewer (current git diff + per-turn diffs)	/diff
/verify	Build & run your app to confirm a change works (not just tests)	/verify
/run	Launch and drive your project's app to see the change live	/run
/run-skill-generator	Write a per-project skill teaching /run and /verify how to launch your app	/run-skill-generator

3. Slash Commands — Planning, Model & Modes

Command	What it does	Example
/plan [description]	Enter plan mode (think first, then propose)	/plan fix the login bug
/model [model]	Switch model and save as default (arrows adjust effort)	/model opus
/effort [level]	Set reasoning effort: low→xhigh, max, ultracode, or auto	/effort high
/fast [on\|off]	Toggle fast mode (skip non-essential work)	/fast on
/advisor [model\|off]	Enable a second model that advises at key moments (fable, opus, sonnet)	/advisor opus
/autocompact [auto\|<tokens>]	Set when context auto-compacts (e.g. 500k)	/autocompact 500k
/permissions	Manage allow/ask/deny tool-permission rules	/permissions
/sandbox	Toggle sandbox mode (isolated execution)	/sandbox
/tui [default\|fullscreen]	Switch terminal renderer (fullscreen = flicker-free)	/tui fullscreen
/theme	Change color theme (auto/light/dark/colorblind-safe)	/theme dark
/color [color]	Set the prompt-bar color (red, blue, green, yellow, purple, orange, pink, cyan)	/color cyan
/statusline	Configure the status line (describe what you want)	/statusline show git branch
/voice [hold\|tap\|off]	Toggle voice dictation	/voice tap

4. Slash Commands — Configuration & Admin

Command	What it does	Example
/init	Create a CLAUDE.md project memory file (interactive walkthrough optional)	/init
/memory	Edit CLAUDE.md files, toggle auto memory, view entries	/memory
/config [key=value...]	Open Settings UI, or set values directly (v2.1.181+)	/config theme=dark
/keybindings	Open your keyboard shortcuts file	/keybindings
/hooks	View hook configurations for tool events	/hooks
/ide	Manage IDE integrations and status	/ide
/mcp [subcommand]	Manage MCP servers + OAuth (no args = interactive list)	/mcp reconnect github
/plugin [subcommand]	Manage plugins (list, install, enable, disable)	/plugin install code-review
/reload-plugins [--force]	Reload all active plugins without restarting	/reload-plugins
/reload-skills	Re-scan skill/command dirs without restarting	/reload-skills
/skills	List available skills (filter, sort by tokens)	/skills
/import [codex\|gemini]	Import config from Codex / Gemini CLI (aliases incl. --dry-run)	/import codex --dry-run
/status	Settings → Status tab: version, model, account, connectivity	/status
/usage	Session cost, plan limits, activity stats (aliases: /cost, /stats)	/usage
/usage-credits	Configure / request usage credits when hitting a limit	/usage-credits
/rate-limit-options	Options when a usage limit blocks a request (wait, credits, upgrade)	/rate-limit-options
/login	Sign in to your Anthropic account	/login
/logout	Sign out	/logout
/privacy-settings	View/update privacy settings (Pro/Max plans)	/privacy-settings
/upgrade	Open the upgrade page for a higher plan tier	/upgrade
/terminal-setup	Install Shift+Enter newline binding for your terminal/editor	/terminal-setup
/setup-bedrock	Configure Amazon Bedrock auth/region/model (hidden unless CLAUDE_CODE_USE_BEDROCK=1)	/setup-bedrock
/setup-vertex	Configure Google Cloud Agent Platform auth/region/model	/setup-vertex
/web-setup	Connect GitHub to Claude Code on the web via local gh CLI	/web-setup
/install-github-app	Install the Claude GitHub App (+ optional Actions workflow)	/install-github-app
/install-slack-app	Install the Claude Slack app (browser OAuth)	/install-slack-app
/auto-mode-setup	Draft autoMode.environment entries from project + sessions	/auto-mode-setup

5. Slash Commands — Agents, Parallel Work & Automation

Command	What it does	Example
/agents	Reminder UI to create/manage subagents (edit .claude/agents/)	/agents
/list-agents	List subagents, teammates, and sessions Claude can message	/list-agents
/subtask <task>	Spawn a forked subagent (inherits full conversation, works in background)	/subtask add unit tests for auth
/tasks	View/manage background work + finished subagents (alias: /bashes)	/tasks
/batch <instruction>	Decompose large change into 5–30 units; run each in its own worktree + PR	/batch migrate src/ from JS to TS
/deep-research <question>	Fan out web searches, cross-check sources, synthesize a cited report	/deep-research best RAG stack for SMB docs
/workflows	Watch, pause, resume, save running/completed workflows	/workflows
/schedule [description]	Create/update/list routines that execute in the cloud	/schedule nightly dependency bump
/loop [interval] [prompt]	Run a prompt repeatedly while the session stays open	/loop 30m run the test suite
/autofix-pr [prompt]	Cloud session watches the current PR and pushes fixes on CI/review failures	/autofix-pr only fix lint errors
/stop	Stop the current background session (keeps transcript + worktree)	/stop

6. Slash Commands — Skills & Extras

Command	What it does	Example
/help	Show help and all available commands	/help
/doctor	Setup checkup: diagnose + fix installation issues	/doctor
/debug [description]	Enable debug logging and troubleshoot from the session log	/debug
/bug [report]	Report a bug / share conversation (consent screen first)	/bug
/feedback [report]	Send product feedback	/feedback
/insights	HTML report analyzing your recent sessions (projects, habits, failures)	/insights
/team-onboarding	Generate a team onboarding guide from your usage history (30 days)	/team-onboarding
/dataviz [request]	Chart/graph design guidance (form, colors, palettes)	/dataviz revenue by quarter
/design [brief]	Draft UI mockups/landing pages as artifact artboards	/design landing page for AI audit tool
/design-login	Authorize design-system access for /design-sync	/design-login
/design-sync [hint]	Upload your React design system so Claude Design uses your components	/design-sync
/artifacts	List/attach/share your artifacts	/artifacts
/chrome	Configure Claude in Chrome settings	/chrome
/desktop	Continue the session in the Claude Code Desktop app (alias: /app)	/desktop
/mobile	QR code for the Claude mobile app (aliases: /ios, /android)	/mobile
/claude-api [action]	Load Claude API / Managed Agents reference for your project's language	/claude-api cost-optimize
/fewer-permission-prompts	Scan transcripts → prioritized allowlist in project settings	/fewer-permission-prompts
/workflow-authoring	Load the dynamic-workflow script reference	/workflow-authoring
/powerup	Interactive lessons on Claude Code features	/powerup
/release-notes	Changelog in an interactive version picker	/release-notes
/passes	Share a free week of Claude Code with friends	/passes
/radio	Open Claude FM lo-fi radio in the browser	/radio
/stickers	Order Claude Code stickers	/stickers

Removed/deprecated: /vim (removed v2.1.92 → use /config → Editor mode), /pr-comments (removed v2.1.91 → just ask Claude), /ultraplan (removed → use plan mode), /enable-auto-mode flag (removed v2.1.111 → auto mode is in the Shift+Tab cycle).

7. CLI Subcommands (terminal)

Command	What it does	Example
claude	Start interactive session	claude
claude "query"	Start session with initial prompt	claude "explain this project"
claude -p "query"	Print mode: one-shot, exits when done (SDK-style)	claude -p "review this diff" --max-turns 10
cat file \| claude -p "query"	Process piped content	cat logs.txt \| claude -p "explain"
claude -c	Continue most recent conversation in this directory	claude -c -p "check for type errors"
claude -r "<session>"	Resume session by ID or name	claude -r "auth-refactor" "finish this PR"
claude update	Update to the latest version	claude update
claude install [version]	Install/reinstall native binary (stable, latest, or version)	claude install stable
claude auth login	Sign in (--email, --sso, --console for API billing)	claude auth login --console
claude auth logout	Sign out	claude auth logout
claude auth status	Show auth status (JSON; --text human-readable; exit 0/1)	claude auth status
claude agents	Agent view: monitor/dispatch background sessions (--json)	claude agents --json
claude attach <id>	Attach to a background session in this terminal	claude attach 7c5dcf5d
claude logs <id>	Print recent output from a background session	claude logs 7c5dcf5d
claude stop <id>	Stop a background session (alias: claude kill)	claude stop 7c5dcf5d
claude respawn <id>	Restart a background session with conversation intact	claude respawn 7c5dcf5d
claude rm <id>	Remove a background session from the list	claude rm 7c5dcf5d
claude mcp	Configure MCP servers (add/list/remove/login/logout)	claude mcp add github -- npx @modelcontextprotocol/server-github
claude plugin	Manage plugins (alias: claude plugins)	claude plugin install code-review@claude-plugins-official
claude doctor	Installation/settings diagnostics (read-only)	claude doctor
claude daemon status	Background-session supervisor state	claude daemon status
claude import [codex\|gemini]	Import config from other coding agents	claude import codex --dry-run
claude setup-token	Generate a long-lived OAuth token for CI (subscription required)	claude setup-token
claude remote-control	Start Remote Control server (control from claude.ai / app)	claude remote-control --name "My Project"
claude gateway --config	Self-hosted Claude apps gateway (Bedrock / Vertex / Foundry, v2.1.195+)	claude gateway --config gateway.yaml
claude project purge [path]	Delete all local Claude Code state for a project	claude project purge ~/work/repo --dry-run
claude ultrareview [target]	Non-interactive deep review (--json, --post to PR)	claude ultrareview 1234 --json
claude auto-mode defaults	Print built-in auto-mode classifier rules (JSON)	claude auto-mode defaults
claude self-hosted-runner setup	Register this machine as a self-hosted runner (v2.1.224+)	claude self-hosted-runner setup

8. Key CLI Flags

Flag	What it does	Example
-p, --print	Non-interactive one-shot mode	claude -p "fix the bug"
-c, --continue / -r, --resume <id>	Continue / resume a session	claude -r auth-refactor
--fork-session	Resume with a new session ID (keeps history)	claude -r abc123 --fork-session
--model <alias>	Model: sonnet, opus, haiku, fable, or full name	claude --model opus
--effort <level>	Reasoning depth: low / medium / high / xhigh / max / ultracode	claude --effort high
--permission-mode <mode>	default, acceptEdits, plan, auto, dontAsk, bypassPermissions, manual	claude --permission-mode plan
--dangerously-skip-permissions	Auto-approve ALL tool use	claude --dangerously-skip-permissions -p "task"
--allowedTools <tools>	Whitelist tools (Bash(git commit *), mcp__server__tool, …)	claude --allowedTools "Read,Edit"
--disallowedTools <tools>	Blacklist tools	claude --disallowedTools "Bash(rm *)"
--tools <tools>	Override the built-in tool set ("" = none, "default" = all)	claude --tools "Bash,Edit,Read"
--max-turns <n>	Cap agentic turns (print mode)	claude -p "task" --max-turns 10
--max-budget-usd <n>	Cap API spend (print mode; min ~$0.05)	claude -p "task" --max-budget-usd 5
--fallback-model <m>	Auto-fallback when primary model is overloaded (chain with commas)	claude -p "task" --fallback-model sonnet,haiku
--output-format <fmt>	text / json / stream-json (print mode)	claude -p "task" --output-format json
--json-schema <schema>	Force validated structured JSON output	claude -p "extract funcs" --json-schema '{...}'
--input-format <fmt>	text / stream-json for stdin	claude -p "task" --input-format stream-json
--verbose	Full turn-by-turn output	claude -p "task" --verbose
--include-partial-messages	Include partial streaming chunks (stream-json)	claude -p "task" --output-format stream-json --include-partial-messages
--append-system-prompt <text>	Append to the default system prompt	claude --append-system-prompt "Always use TypeScript"
--system-prompt <text>	Replace the entire system prompt	claude --system-prompt "You are a Python expert"
--bare	Minimal mode: skip hooks, skills, plugins, MCP, CLAUDE.md, OAuth	claude --bare -p "task"
--safe-mode	All customizations disabled (troubleshoot config; keeps auth/tools)	claude --safe-mode
--restricted	Restricted mode for eval harnesses on shared machines (v2.1.248+)	claude --restricted -p "task"
--agents '<json>'	Define custom subagents dynamically	claude --agents '{"reviewer":{"description":"Reviews code","prompt":"You are a code reviewer"}}'
--mcp-config <path> / --strict-mcp-config	Load MCP servers from file; ignore all other MCP configs	claude --mcp-config ./mcp.json --strict-mcp-config
--add-dir <paths>	Grant additional working directories	claude --add-dir ../apps ../lib
-w, --worktree [name]	Run in an isolated git worktree (.claude/worktrees/<name>)	claude -w feature-auth
--tmux	Create a tmux session for the worktree	claude -w feature-auth --tmux
--bg, --background	Start as a background agent, return immediately	claude --bg "investigate the flaky test"
--exec	Run a shell command as a PTY background job (with --bg)	claude --bg --exec 'pytest -x'
--cloud	Create a web session on claude.ai (or queue into existing)	claude --cloud "Fix the login bug"
--from-pr <number>	Session picker filtered to sessions linked to a PR	claude --from-pr 123
--name <name>	Display name for the session (resumable)	claude -n "my-feature-work"
--session-id <uuid>	Use a specific session UUID	claude --session-id "550e8400-..."
--no-session-persistence	Don't save session to disk (print mode; CI)	claude -p "task" --no-session-persistence
--autocompact <auto\|tokens>	Set auto-compact window (v2.1.221+)	claude --autocompact 500k
--debug[=filter] / --debug-file <path>	Debug logging (with category filter)	claude --debug='mcp,startup'
--ide	Auto-connect to the IDE on startup	claude --ide
--chrome / --no-chrome	Enable/disable Chrome browser integration	claude --chrome
--version	Print the version	claude -v

9. Keyboard Shortcuts & Input Prefixes

Shortcut	Action
Ctrl+C	Interrupt / clear input (2nd press exits)
Ctrl+D	Exit session
Ctrl+R	Reverse-search command history
Ctrl+B	Background a running task
Ctrl+V	Paste image into the conversation
Ctrl+O	Transcript mode — see Claude's thinking
Ctrl+G / Ctrl+X Ctrl+E	Open prompt in external editor
Esc Esc	Rewind conversation / code state (or summarize)
Shift+Tab	Cycle permission modes (Normal → Auto-Accept → Plan → Auto)
Alt+P	Switch model
Alt+T	Toggle thinking mode
Alt+O	Toggle Fast Mode
Shift+Enter	Newline
?	Shortcuts help (fullscreen renderer)

Prefix	Action	Example
/	Slash commands	/compact
!	Run bash directly, bypassing the AI (! alone toggles shell mode)	!npm test
@	Reference files/dirs with autocomplete	@./src/api/
#	Quick add to CLAUDE.md memory	# Use 2-space indentation
\ + Enter	Quick newline	—
"ultrathink"	Max reasoning on this turn, regardless of /effort	ultrathink: analyze this algorithm

Sources

• Claude Code docs — CLI reference (commands, flags)
• Claude Code docs — Commands (full slash-command table, 111 commands)
• Claude Code docs — Interactive mode (shortcuts, input modes)