GTM Resource Hub

Claude Code CLI Cheatsheet: Essential Commands and Shortcuts

You are using 20% of what Claude Code can do. This cheatsheet covers the commands, flags, and shortcuts that separate casual usage from real output.

Claude Code is Anthropic's agentic coding assistant that runs directly in your terminal. It reads files, executes commands, edits code, and reasons about your entire codebase -- no copy-paste, no IDE plugins, no web interface required.

This cheatsheet covers every keyboard shortcut, slash command, and CLI flag you need. Everything here is verified against the official Claude Code documentation as of April 2026. Bookmark it, print it, or keep it open in a second tab.

Most Used Claude Code CLI Commands (Start Here)

These twelve shortcuts and commands show up in nearly every Claude Code session. If you learn nothing else from this cheatsheet, commit these to muscle memory.

Quick Reference -- Most Used

ShortcutDescription
Ctrl+CCancel current input or generation
Ctrl+DExit Claude Code session (EOF)
TabAutocomplete file paths in prompts
Up/DownNavigate command history
/helpShow all available commands
/clearClear conversation, free context
/compactSummarize context to save tokens
Shift+EnterNew line without submitting
Esc + EscRewind to a checkpoint
/exitExit session (alias: /quit)
Ctrl+RReverse search command history
Shift+TabCycle permission modes

Setup & Installation

Claude Code installs via npm and runs anywhere Node.js does. Zero to first session in under five minutes.

How to Install Claude Code CLI

  1. Install Node.js 18+

    Claude Code requires Node.js version 18 or higher. Check with node --version. Download from nodejs.org if needed.

  2. Install Claude Code globally

    Add it to your system path:

    npm install -g @anthropic-ai/claude-code
  3. Authenticate

    You have two options. For a Claude subscription, run claude auth login. For API billing, run claude auth login --console. Or set your key directly:

    export ANTHROPIC_API_KEY=your_key_here
  4. Navigate to your project

    Claude Code works best when launched inside your codebase:

    cd /path/to/your/project
  5. Launch Claude Code

    Start an interactive session:

    claude
Pro Tip

Add export ANTHROPIC_API_KEY=your_key to your ~/.zshrc or ~/.bashrc so you never have to set it again. Run source ~/.zshrc after saving. Alternatively, you can set environment variables permanently in your settings.json.

Session Management Keyboard Shortcuts

These shortcuts control your Claude Code session -- starting, stopping, pausing, and managing the conversation flow. All shortcuts verified against the official interactive mode docs.

Shortcut Action Notes
Ctrl+C Cancel current input or generation Standard interrupt. Stops a running task or clears pending input
Ctrl+D Exit session EOF signal; cleanly closes the session and returns to your shell
Ctrl+L Clear prompt input Clears typed text but preserves conversation history
Esc+Esc Rewind or summarize Restore code and/or conversation to a previous checkpoint. Also available via /rewind
Shift+Tab Cycle permission modes Cycles through default, acceptEdits, plan, and any enabled modes (auto, bypassPermissions)
Alt+T Toggle extended thinking Enable deeper reasoning mode. Mac: Option+T (requires terminal Meta config)
Alt+P Switch model Change models (Sonnet, Opus, Haiku) without clearing your prompt. Mac: Option+P
Alt+O Toggle fast mode Enable faster responses with the same model. Mac: Option+O
Shift+Enter Insert newline without submitting Works out of the box in iTerm2, WezTerm, Ghostty, Kitty. Run /terminal-setup for other terminals
Enter Submit prompt Sends your current input to Claude
Ctrl+T Toggle task list Show or hide the task list in the terminal status area
Ctrl+O Toggle transcript viewer Cycles through normal, transcript mode, and focus view. Shows detailed tool usage
Ctrl+G Open in external editor Edit prompts in your default text editor. Also: Ctrl+X Ctrl+E
Ctrl+B Background running task Send bash commands and agents to the background. Tmux users: press twice
Ctrl+X Ctrl+K Kill all background agents Press the sequence twice within 3 seconds to confirm
Ctrl+V Paste image from clipboard Inserts an [Image #N] chip. Mac in iTerm2: Cmd+V. Windows: Alt+V
Hold Space Push-to-talk dictation Requires /voice to be enabled. Transcript inserts at cursor
macOS Note

Option/Alt shortcuts (Alt+T, Alt+P, Alt+O) require configuring Option as Meta in your terminal. In iTerm2: Settings > Profiles > Keys > set Left/Right Option to "Esc+". In VS Code: set "terminal.integrated.macOptionIsMeta": true.

Navigate command history, search previous prompts, and use the quick-command prefixes that make Claude Code faster than a chat window.

Shortcut Action Notes
/ Navigate command history Recall previous inputs. History is per working directory and resets on /clear
Ctrl+R Reverse search command history Type to search. Ctrl+R again to cycle matches. Tab/Esc to accept, Enter to execute
Tab Autocomplete file path or command Works with @ file mentions and after ! bash prefix
@ File path mention Triggers file path autocomplete. E.g., @src/auth.ts fix the bug
! at start Bash mode Run shell commands directly. Output added to conversation context. E.g., ! npm test
/ at start Command or skill Type / and keep typing to filter available commands
Alt+B Move cursor back one word Word navigation. Mac: requires Option as Meta
Alt+F Move cursor forward one word Word navigation. Mac: requires Option as Meta
/ Cycle dialog tabs Navigate between tabs in permission dialogs and menus

Multiline Input Methods

Claude Code supports several ways to write multi-line prompts.

Method How Notes
Backslash escape \ then Enter Works in all terminals
macOS default Option+Enter Default on macOS
Shift+Enter Shift+Enter Works in iTerm2, WezTerm, Ghostty, Kitty. Run /terminal-setup for others
Control sequence Ctrl+J Line feed character. Works everywhere
Paste mode Paste directly For code blocks, logs, and large text

Text Editing Shortcuts

Claude Code uses readline-style editing. These shortcuts let you cut, paste, and manipulate prompt text without reaching for the mouse.

Shortcut Action Notes
Ctrl+K Delete to end of line Stores deleted text for pasting with Ctrl+Y
Ctrl+U Delete from cursor to line start Stores deleted text for pasting. Repeat to clear across lines in multiline input
Ctrl+Y Paste deleted text (yank) Pastes text deleted with Ctrl+K or Ctrl+U. Different from system clipboard
Alt+Y Cycle paste history After Ctrl+Y, cycle through previously deleted text. Mac: requires Option as Meta
Vim Mode Available

Prefer vim-style editing? Enable it via /config > Editor mode. You get full normal mode navigation (h/j/k/l, w/b/e), editing operators (d, c, y, p), and text objects (iw, i", i{). Full vim reference in the docs.

All Slash Commands

Type / in Claude Code to see every available command, or type / followed by letters to filter. Commands marked [Skill] are bundled skills -- prompts handed to Claude that it can also invoke automatically when relevant. Full list at the official commands reference.

Session Control

Command Description When to Use
/help Show all available commands First thing to run in a new session
/clear Clear conversation history and free context When switching tasks. Aliases: /reset, /new
/compact [instructions] Summarize conversation to reclaim context space When context fills up in long sessions. Pass instructions to guide the summary
/exit Exit Claude Code and return to the shell Alias: /quit
/status Show session info (model, context usage, cost) Monitor token consumption during long sessions
/cost Show cumulative cost for this session Track spending, especially in automation
/model [model] Display or switch the active model Use sonnet or opus aliases, or full model IDs
/effort [level] Set effort level (low, medium, high, max) Lower effort for simple tasks; max for Opus 4.6 deep reasoning
/fast [on|off] Toggle fast mode Same model, faster output

Context & Navigation

Command Description When to Use
/resume [session] Resume a conversation by ID or name Pick up where you left off. Alias: /continue
/rename [name] Set a display name for the current session Shown in /resume picker and prompt bar
/branch [name] Branch the conversation at this point Try different approaches without losing history. Alias: /fork
/rewind Rewind conversation and/or code to a checkpoint Supports code-only or conversation-only rollback. Alias: /checkpoint
/diff Show all changes made this session Review before committing
/copy [N] Copy last response to clipboard Pass N for Nth-latest. Code block picker available. w writes to file
/export [filename] Export conversation to file Save the full session transcript
/btw <question> Ask a side question without adding to history Quick answers that do not clutter context. Works while Claude is processing
/context Show current context window usage See how full the context is
/memory Manage CLAUDE.md memory files Edit persistent project instructions

Configuration & Setup

Command Description When to Use
/config Open settings interface Adjust theme, model, output style, editor mode. Alias: /settings
/permissions Manage tool permission rules View/edit allow, ask, and deny rules by scope. Alias: /allowed-tools
/init Initialize project with CLAUDE.md Run in every new repository to create persistent project context
/keybindings Open keyboard shortcuts config Customize at ~/.claude/keybindings.json
/doctor Run diagnostics Debug keybinding conflicts, MCP issues, plugin problems
/theme Change color theme Customize Claude Code appearance
/terminal-setup Install terminal keybindings Needed for Shift+Enter in VS Code, Alacritty, Zed, Warp
/mcp Manage MCP server connections Add, remove, or configure Model Context Protocol servers
/hooks View configured hooks See active lifecycle hooks for the session
/login / /logout Manage authentication Sign in or out of your Anthropic account

Advanced Skills & Automation

These are bundled skills -- prompt-based workflows that Claude executes as multi-step tasks.

Command Description Example
/batch <instruction> Orchestrate large-scale changes in parallel /batch migrate src/ from Solid to React
/simplify [focus] Review recent changes for quality and efficiency, then fix /simplify focus on memory efficiency
/autofix-pr [prompt] Watch current PR and push fixes for CI failures and reviews /autofix-pr only fix lint errors
/loop [interval] <prompt> Run a prompt repeatedly on an interval /loop 5m check if the deploy finished
/schedule [description] Create scheduled tasks Set up tasks to run at specific times
/ultraplan <prompt> Deep planning for complex multi-step tasks /ultraplan redesign the auth system
/plan [description] Enter plan mode directly /plan fix the auth bug
/debug [description] Enable debug logging and troubleshoot issues /debug MCP server not connecting
/security-review Security audit of recent changes Run before pushing sensitive code changes

Remote & Cross-Device

Command Description Notes
/remote-control Make this session controllable from claude.ai Alias: /rc. Control from web or mobile app
/teleport Pull a web session into this terminal Alias: /tp. Fetches branch and conversation
/desktop Continue session in Claude Code Desktop app macOS and Windows only. Alias: /app
/voice Enable push-to-talk voice input Hold Space to dictate after enabling
/chrome Enable Chrome browser integration For web automation and testing
/ide Connect to IDE Auto-connect to a running IDE instance

CLI Flags & Startup Options

These flags are passed when launching Claude Code from the command line. Note: claude --help does not list every flag -- a flag's absence from --help does not mean it is unavailable. Full reference at code.claude.com/docs/en/cli-reference.

Core Commands

Command Description Example
claude Start interactive session claude
claude "query" Start session with initial prompt claude "explain this project"
claude -p "query" Print mode: run prompt, then exit claude -p "explain this function"
claude -c Continue most recent conversation claude -c
claude -r <session> "query" Resume session by ID or name claude -r "auth-refactor" "Finish this PR"
claude update Update to latest version claude update
claude auth login Sign in to Anthropic account claude auth login --console for API billing
claude agents List all configured subagents claude agents
claude mcp Configure MCP servers claude mcp add my-server
claude plugin Manage plugins claude plugin install code-review@claude-plugins-official
claude remote-control Start Remote Control server mode claude remote-control --name "My Project"
cat file | claude -p "query" Pipe content as input cat logs.txt | claude -p "explain these errors"

Session & Model Flags

Flag Description Example
--model Set the model. Accepts aliases (sonnet, opus) or full IDs claude --model sonnet
--effort Set effort level: low, medium, high, max claude --effort high
--continue / -c Load most recent conversation claude -c
--resume / -r Resume session by ID or name claude -r auth-refactor
--name / -n Set session display name claude -n "my-feature-work"
--fork-session Create new session ID when resuming claude --resume abc123 --fork-session
--fallback-model Auto-fallback model when primary is overloaded claude -p --fallback-model sonnet "query"

Permissions & Tools Flags

Flag Description Example
--permission-mode Start in a specific permission mode (default, acceptEdits, plan, auto, bypassPermissions) claude --permission-mode plan
--dangerously-skip-permissions Skip all permission prompts Equivalent to --permission-mode bypassPermissions
--allowedTools Tools that execute without prompting claude --allowedTools "Bash(git log *)" "Read"
--disallowedTools Remove specific tools entirely claude --disallowedTools "Bash(git push *)"
--tools Restrict which built-in tools Claude can use claude --tools "Bash,Edit,Read" or "" for none

Automation & CI Flags

Flag Description Example
--output-format Output format for print mode: text, json, stream-json claude -p "list files" --output-format json
--max-turns Limit agentic turns in print mode claude --max-turns 10 -p "fix tests"
--max-budget-usd Maximum dollar spend before stopping claude -p --max-budget-usd 5.00 "query"
--json-schema Get validated JSON matching a schema claude -p --json-schema '{"type":"object",...}' "query"
--bare Minimal mode: skip hooks, skills, plugins, MCP, CLAUDE.md claude --bare -p "query"
--verbose Full turn-by-turn logging claude --verbose
--debug Debug output with optional category filter claude --debug "api,mcp"

Remote & Worktree Flags

Flag Description Example
--worktree / -w Start in isolated git worktree claude -w feature-auth
--tmux Create tmux session for worktree. Requires -w claude -w feature-auth --tmux
--remote Create a web session on claude.ai claude --remote "Fix the login bug"
--remote-control / --rc Start interactive session with Remote Control claude --rc "My Project"
--teleport Resume a web session in your terminal claude --teleport
--chrome Enable Chrome browser integration claude --chrome
--add-dir Add extra working directories claude --add-dir ../apps ../lib

System Prompt Flags

Flag Description Example
--system-prompt Replace the entire system prompt claude --system-prompt "You are a Python expert"
--system-prompt-file Replace prompt from file claude --system-prompt-file ./review.txt
--append-system-prompt Append to default prompt claude --append-system-prompt "Always use TypeScript"
--append-system-prompt-file Append file contents to default prompt claude --append-system-prompt-file ./rules.txt
Use --dangerously-skip-permissions Carefully

This flag (and the equivalent --permission-mode bypassPermissions) auto-approves every operation including file deletions and shell commands. It is designed for CI pipelines and automation scripts where you trust the environment -- not for interactive development where you want to review changes first.

Pro Tips & Productivity Patterns

These patterns come from real usage across teams building with Claude Code daily. None of them are obvious from the docs alone.

1. Use /compact before you hit the wall

Claude Code has a context window limit. In sessions where you have explored dozens of files, the model starts losing track of early context. Run /compact proactively -- every 30 minutes or whenever you shift to a new subtask. You can pass instructions to guide the summary: /compact keep the auth refactor context, drop the test exploration.

2. Scope prompts with @ file mentions

Generic prompts like "fix the authentication bug" force Claude to search your entire codebase. Reference specific files with @src/auth/session.ts syntax to reduce hallucination and speed up responses. Tab-completion works with @, so you do not need to type full paths.

3. Use /btw for quick questions mid-task

When Claude is in the middle of a long task, use /btw to ask a side question without polluting conversation context. The answer appears in an overlay and never enters history. Useful for "what was that config file called again?" moments.

4. Match model to task with /model

Not every task needs Opus. Switch to /model haiku for quick tasks -- renaming variables, generating docstrings, writing utility functions. Save Opus for architectural reasoning, complex debugging, and multi-file refactors. The cost difference is 10-20x at scale.

5. Chain --output-format json with pipes for automation

When using Claude Code in scripts and CI pipelines, combine -p with --output-format json for machine-parseable output. Add --max-budget-usd to cap spend: claude -p "analyze test failures" --output-format json --max-budget-usd 2.00

6. Pin a CLAUDE.md for persistent project context

Create a CLAUDE.md file in your project root. Claude Code reads it at the start of every session. Use it for code style preferences, architectural decisions, naming conventions, and domain knowledge. Run /init to generate one automatically, or /memory to edit it mid-session.

7. Use worktrees for parallel sessions

Working on multiple features? Use --worktree to isolate each session in its own git worktree: claude -w feature-auth. This creates a worktree at .claude/worktrees/feature-auth. Combine with --tmux to get each session in its own pane.

8. Remote Control for mobile monitoring

Start a session with --rc to control it from claude.ai or the Claude mobile app. Start a long-running task, walk away from your desk, and check progress from your phone. Combine with worktrees for isolated remote sessions: claude -w bugfix --rc "Fix the login bug"

9. Use /batch for large-scale refactors

The /batch skill decomposes large changes into 5-30 independent units, then spawns parallel agents in isolated worktrees. Each agent implements its unit, runs tests, and opens a PR. This turns a day of manual refactoring into a 30-minute supervised operation.

10. Save shell aliases for repeated workflows

If you run the same Claude Code prompts repeatedly, save them as shell aliases in your ~/.zshrc:

Alias Expands To
cc-review claude -p "review the staged changes for bugs, security issues, and style"
cc-tests claude -p "generate unit tests for modified files in this branch"
cc-commit claude -p "write a conventional commit message for my staged changes"
cc-explain claude -p "explain this codebase for a new engineer"
GU

Guest

Writer at Octave

Build your generative GTM motion today

Placeholder Image