claude-refcard.md

Claude Code Cheat Sheet

Everything you need in one place — Commands, Shortcuts, Features & Tips

2026 EDITION

---

⌨️ Keyboard Shortcuts

Essential

Key	Action
Enter	Send message / submit
Esc	Interrupt / stop generation
Esc Esc	Open rewind menu (go back in conversation or code)
Ctrl+C	Cancel current operation (hard stop)
Ctrl+D	Exit Claude Code
Shift+Tab	Cycle modes: Normal → Auto-Accept → Plan

Navigation

Key	Action
Ctrl+R	Search command history
Ctrl+T	Toggle task list
Ctrl+O	Toggle verbose transcript
Ctrl+G	Open external editor (write long prompts)
Ctrl+V	Paste image (screenshots, diagrams)
Ctrl+S	Stash current prompt (save for later)
Cmd+P / Meta+P	Open model picker (switch models quick)
Cmd+T / Meta+T	Toggle extended thinking

Editing (Bash-style)

Key	Action
Ctrl+A / Ctrl+E	Start / end of line
Opt+F / Opt+B	Word forward / back
Ctrl+W	Delete previous word
\ + Enter	New line (without sending)

Background Tasks

Key	Action
Ctrl+B	Send running task to background

💡 Tip: Run /terminal-setup to enable Shift+Enter for multi-line input in iTerm2 & VS Code. Run /keybindings to customize all shortcuts.

---

⚡ Slash Commands

Session Control

Command	Action
/clear	Reset conversation history (fresh start)
/compact [hint]	Compress context to save tokens. Optional hint for what to keep.
/rewind	Go back in conversation AND/OR code changes
/export [file]	Export conversation to file or clipboard
/cost	Show session cost & token usage
/usage	Show plan usage & rate limits
/context	Token consumption visualization

Configuration

Command	Action
/config	Open settings panel
/model	Switch between Sonnet / Opus / Haiku
/permissions	View & update tool permissions
/keybindings	Open keyboard shortcuts config file
/vim	Toggle vim mode for input
/terminal-setup	Setup Shift+Enter for multi-line input

Development

Command	Action
/init	Create CLAUDE.md for your project — do this first!
/memory	View & edit CLAUDE.md project memory
/review	Code review analysis
/doctor	Environment diagnostics & health check
/agents	Manage sub-agents
/mcp	Manage MCP servers

Advanced

Command	Action
/insights	Generate HTML usage report 🆕
/pr_comments	View GitHub PR feedback
/install-github-app	Setup automated PR reviews
/tasks	Persistent task list management
/teleport	Transfer session between web ↔ local

---

🚀 CLI Launch Flags

Starting Sessions

Flag	Action
claude	Start interactive session
claude "query"	Start with an initial prompt
claude -p "query"	Print mode — answer & exit (for scripting)
claude -c	Continue last conversation
claude -r "name"	Resume specific session by name or ID
claude -w name	Start in isolated git worktree

Model & Behavior

Flag	Action
--model sonnet	Use Sonnet (fast, cheap)
--model opus	Use Opus (smartest)
--agent my-agent	Use a specific sub-agent
--permission-mode plan	Start in plan mode
--max-turns N	Limit conversation turns
--max-budget-usd N	Set max spend limit

Context & Directories

Flag	Action
--add-dir ../path	Add extra directories to context
--chrome	Enable browser integration
--verbose	Show detailed logging

Permissions

Flag	Action
--allowedTools	Whitelist specific tools
--disallowedTools	Block specific tools
--tools "Bash,Edit"	Restrict to only these tools

Output Formats (for -p mode)

Flag	Action
--output-format text	Plain text (default)
--output-format json	Structured JSON
--output-format stream-json	Real-time streaming JSON

💡 Tip: Pipe data in! git diff | claude -p "review this" or cat error.log | claude -p "explain"

---

🧩 The Big 5 — Claude Code Extension System

1. CLAUDE.md — Project Memory

What	A markdown file Claude reads every session. Your project's "brain dump" — coding style, architecture, common commands, conventions.
Where	.claude/CLAUDE.md (project) or ~/.claude/CLAUDE.md (global)
Create	Run /init in your project — Claude generates it for you

2. Custom Slash Commands

What	Your own /commands. Markdown files with prompts that YOU invoke. Like prompt templates.
Where	.claude/commands/ (project) or ~/.claude/commands/ (global)
Use	Filename = command name. review.md → type /project:review

3. Skills — Auto-Invoked Knowledge

What	Like commands, but Claude decides when to use them automatically. You DON'T invoke them — Claude detects when they're relevant.
Where	.claude/skills/ with a SKILL.md inside each skill folder
Use	Just work on your project — Claude picks up relevant skills from context

4. Sub-Agents — Specialized Helpers

What	Separate Claude instances with their own context & role. Like team members: reviewer, debugger, architect, etc.
Where	.claude/agents/ (markdown files with YAML metadata)
Invoke	/agents to manage, or just say "Use the reviewer agent"
CLI	--agent my-agent or --agents '{json}'

5. MCP Servers — External Tool Connections

What	Connect Claude to external tools: GitHub, Notion, databases, APIs, browsers, etc.
Setup	claude mcp add <name> <command>
List	claude mcp list
Config	--mcp-config ./mcp.json at launch

+ Plugins — Community Extensions

What	Bundles of commands, skills, hooks & more from the community
Browse	/plugin to browse, install, enable, disable
Dir	--plugin-dir ./my-plugins for local plugins

How they differ:

Custom Commands → YOU invoke them vs Skills → CLAUDE invokes them vs Sub-Agents → Separate AI instances vs MCP → External tool connections

---

🔄 Permission Modes

Mode	Description
Normal	Claude asks permission for every tool use (read, write, bash, etc.)
Auto-Accept	Claude runs tools WITHOUT asking. Faster but less control. Good for trusted tasks.
Plan Mode	Claude ONLY reads & plans. Won't write or run anything. Review first, then switch to Normal to execute.

Cycle: Shift+Tab → Normal → Auto-Accept → Plan → Normal...

💡 Best workflow: Start in Plan Mode to explore & understand the problem. Review Claude's plan. Switch to Normal/Auto-Accept to implement.

---

🪝 Hooks — Event Automation

Hook	Description
PreToolUse	Runs BEFORE Claude uses a tool — validate, block, or modify
PostToolUse	Runs AFTER a tool — check results, auto-format, lint
UserPromptSubmit	Before your message is processed
Stop	When Claude finishes its response
SessionStart	When a session begins
SessionEnd	When a session ends
PreCompact	Before context compression
Notification	When Claude sends a notification

💡 Example: Auto-run prettier after every file edit, or block writes to .env files. Configure in your settings JSON.

---

✨ Input Superpowers

Feature	Description
@ mention	Type @ to reference files & folders. Claude reads them into context.
! prefix	Type ! to run shell commands inline. E.g., ! git status
Paste images	Ctrl+V to paste screenshots, diagrams, error images directly
Pipe input	cat file.py | claude -p "explain" — feed data directly
Multi-dir	claude --add-dir ../api ../web — work across multiple projects
Worktrees	claude -w feature — isolated git branch + Claude session

⚠️ Pro tip: Use @ references instead of copy-pasting file contents. It's smarter with context and uses fewer tokens.

---

⚙️ Configuration

Settings Priority (highest → lowest)

Level	Location
Enterprise	/etc/claude-code/managed-settings.json
Project Local	.claude/settings.local.json (your personal project settings)
Project Shared	.claude/settings.json (committed to git, shared with team)
User Global	~/.claude/settings.json (your defaults)

Config CLI

Command	Action
claude config list	Show all settings
claude config get key	Check a value
claude config set key value	Change a value
claude config add key value	Add to array

💡 Permissions example: Allow git commands without asking: add "Bash(git:*)" to your allowedTools in settings.

---

📁 File Structure Map

Project Level (.claude/)

File/Dir	Purpose
CLAUDE.md	Project memory — conventions, architecture, commands
settings.json	Shared project settings (committed to git)
settings.local.json	Your personal settings (gitignored)
commands/	Project slash commands (*.md files)
skills/	Project skills (folders with SKILL.md)
agents/	Project sub-agents (*.md files)

Global Level (~/.claude/)

File/Dir	Purpose
CLAUDE.md	Global memory (applies to ALL projects)
settings.json	Global settings
commands/	Personal global commands
skills/	Personal global skills
keybindings.json	Custom keyboard shortcuts

---

⏪ Rewind & Checkpoints

Key/Command	Action
Esc Esc	Open rewind menu anywhere
/rewind	Same but typed as command

Rewind Options

Option	Description
Conversation	Go back in chat only. Code stays as-is.
Code	Restore files only. Conversation stays.
Full Rewind	Restore both conversation AND code to a point.

⚠️ Note: Bash side-effects (database changes, API calls, deleted files via rm) can't be rewound. Checkpoints only track file edits by Claude. Use Git for permanent safety.

---

🎯 Pro Workflow — How to Get the Best Out of Claude Code

Starting a New Project

cd project && claude → /init → Edit CLAUDE.md → Code!

The Plan → Execute Pattern

Shift+Tab → Plan Mode → Describe what you want → Review Claude's plan → Shift+Tab → Normal/Auto → Execute

Saving Money

Strategy	Why
Use /compact	When context gets big, compress it. Saves tokens dramatically.
Use /clear	Between unrelated tasks. Don't carry irrelevant context.
Use Sonnet	For routine tasks. Save Opus for complex architecture decisions.
Use @ refs	Instead of pasting code — smarter context management.

Debugging Like a Pro

Strategy	Why
Paste errors	Copy-paste the full error message. Claude parses stack traces brilliantly.
Paste screenshots	Ctrl+V a screenshot of the bug. Claude sees it.
Pipe logs	cat error.log | claude -p "what's wrong?"
/doctor	If something feels broken, run this first.

Parallel Development

Strategy	How
Worktrees	claude -w feature-auth — isolated branch + session
Multiple dirs	--add-dir ../api ../web — work across repos
Background	Ctrl+B sends a task to background so you can start another
Agent Teams	Multiple Claude instances collaborating (experimental) 🆕

---

🛠️ Create Custom Commands

Step	Action
1. Create file	.claude/commands/review.md
2. Write prompt	The markdown content IS the prompt Claude will use
3. Use it	Type /project:review in Claude Code

Optional YAML Frontmatter

Key	Purpose
argument-hint	Placeholder text for argument input
description	Shows in /help listing
allowed-tools	Restrict what tools the command can use
model	Force a specific model for this command

Variable: $ARGUMENTS

Use $ARGUMENTS in your markdown — it gets replaced with whatever you type after the command.

💡 Example: /project:review src/auth.ts → $ARGUMENTS = "src/auth.ts"

---

📋 Quick Reference — Most Used Combos

Daily Essentials

What	How
Start project	cd project && claude
Continue where I left off	claude -c
Quick question, no session	claude -p "how do I..."
Review my changes	git diff | claude -p "review"
Explain error	cat error.log | claude -p "explain"
Check cost	Type /cost anytime
Undo mistake	Esc Esc → rewind

Power Moves

What	How
Parallel sessions	claude -w feature-a + claude -w feature-b
Custom reviewer agent	Create .claude/agents/reviewer.md
Auto-format on edit	PostToolUse hook → run prettier
Web session	claude --remote "fix the bug"
Transfer to local	claude --teleport
Budget limit	claude -p --max-budget-usd 2 "query"
Scripted automation	claude -p --output-format json "query" | jq