Claude Code Command Cheat Sheet: Master All CLI Commands and Workflows

Core Value and Design Philosophy

Claude Code (released by Anthropic in 2025) is a terminal‑based AI programming assistant that can read project files, run shell commands, and manipulate Git directly. Its advantage stems from three design pillars:

Multi‑Agent Parallel Architecture – ~40 built‑in tools and a 46 000‑line query engine enable sub‑agents to collaborate on complex tasks.

Progressive Context Management – Hierarchical memory with smart compression keeps context quality high while dramatically reducing token usage.

Deep Repository Operations – Native support for cross‑file edits, Git workflow integration, and fine‑grained permission control.

Installation & Environment Setup

Claude Code requires Node.js 18 or newer. Verify the runtime: node --version Installation scripts for macOS/Linux/WSL:

# Stable version
curl -fsSL https://claude.ai/install.sh | bash
# Latest version
curl -fsSL https://claude.ai/install.sh | bash -s latest
# Homebrew (macOS)
brew install --cask claude-code

Windows (requires Git for Windows or WSL):

# CMD / PowerShell
curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

Optional NPM installation (works on all platforms): npm install -g @anthropic-ai/claude-code After installation, confirm the binary is available: claude --version The first launch walks you through authentication and workspace configuration.

Basic CLI Commands

claude

– Start an interactive REPL in the current directory (daily multi‑turn development). claude "task description" – Launch with a one‑shot question and begin analysis immediately (quick query). claude -p "query" – Execute a single command, print the result, then exit (scripting / CI‑CD). claude -c – Continue the most recent conversation in the current directory (resume after a restart).

Example session:

# Interactive mode
claude
> Please analyze the project’s code structure

# One‑shot query
claude "Explain the authentication flow"

# One‑off execution (append output to a log)
claude -p "Check code style issues" >> review.log

# Resume previous session
claude -c

Advanced Options

--add-dir <paths>

– Add extra work directories (essential for monorepos). Example:

claude --add-dir ../lib ../apps

--model sonnet|opus|haiku

– Select the model for the current session. Example:

claude --model opus

--verbose

– Enable detailed logs of tool execution. Example:

claude --verbose

--dangerously-skip-permissions

– Bypass all permission prompts (use only in sandbox experiments). Example:

claude --dangerously-skip-permissions

Slash Commands (Core Operations)

Session & Context Management

/help

– Show all available commands and shortcuts (when you forget a command). /clear – Fully reset conversation history (switch to a completely different task). /compact – Compress history while retaining a core summary (use when token usage approaches the limit). /context – Visualize current token usage (monitor consumption). /resume – Restore a previous session (multi‑task parallelism). /rewind (ESC × 2) – Undo the last AI‑generated change (quick rollback).

Git & Code Operations

/commit

– Analyze staged changes and auto‑generate a conventional commit message (e.g., feat: add JWT authentication ). /review – Inspect current changes for bugs, style violations, and potential issues (pre‑PR review). /diff – Show the current git diff (understand what changed). /branch – Create or switch Git branches (branch management). /pr_comments – Fetch open PR review comments (process code‑review feedback).

Memory & Configuration

/init

– Generate an initial CLAUDE.md knowledge base (first‑time project onboarding). /memory – Edit the persistent CLAUDE.md file (update project rules or personal preferences). /model – Switch the model mid‑session (Sonnet / Opus / Haiku) based on task complexity. /config – View or modify runtime configuration (tune parameters). /cost – Show token usage and estimated cost for the session (cost control).

Diagnostics & Authentication

/doctor

– Check environment, authentication, and network connectivity (when you encounter connection issues). /login – Log in to an Anthropic account (first use or account switch). /logout – Log out of the current account (switch accounts). /version – Display the installed version (verify you are up‑to‑date). /upgrade – Upgrade to the latest release (gain new features).

Integration & Extension

/mcp

– Manage Model Context Protocol connections to external services (GitHub, databases, etc.). /skills – List and manage reusable skill packages (install custom skills). /hooks – Configure automation hooks (e.g., run prettier after file writes).

Keyboard Shortcuts for Speed

Ctrl+C

– Interrupt the current generation or command (AI goes off‑track). Ctrl+D – Safely exit Claude Code (end session). Ctrl+L – Clear the screen while preserving context (visual clutter). Ctrl+T – Toggle the task list and view background jobs (monitor AI‑created tasks). Ctrl+R – Reverse‑search command history (find previous commands quickly). Shift+Tab – Cycle through the three work modes (Default → Auto‑Accept → Plan). ESC × 2 – Rewind the last AI change (undo). Ctrl+B – Send the current task to the background (e.g., compile while continuing the conversation). Shift+Enter – Insert a newline in the input box without sending (multi‑line prompts). Ctrl+U – Delete the entire input line (quick edit).

Three Work Modes

Default – Every file edit or command requires explicit confirmation (tight control for production).

Auto‑Accept – File modifications are applied automatically; shell commands still need confirmation (rapid iteration when you trust routine actions).

Plan – Read‑only mode; Claude only analyzes and proposes plans (design phase before committing changes).

Switch modes with Shift+Tab.

Advanced Features: Skills, Agents, MCP

Skills (Reusable Toolkits)

Skills are packaged capabilities that can be invoked like plugins. A typical example is a “Run Tests” skill that automatically detects the project structure, runs the test suite, parses the output, and returns a concise failure summary. To create a custom skill, place a SKILL.md file under ~/.claude/skills/.

Agents (Autonomous AI Entities)

Agents can independently plan, execute, reflect, and iterate to achieve a goal. Example workflow: “Fix this GitHub issue” → read the issue description → run tests → locate the bug → generate a fix → open a PR with an explanatory comment.

Model Context Protocol (MCP)

MCP lets Claude Code connect to external services (GitHub, databases, etc.) via the /mcp command. In practice, replace MCP with CLI + Skills when possible to save token budget.

Pros & Cons

Advantages

Strong multi‑agent capabilities – 67 % win rate over Cursor in code‑quality benchmarks.

Cross‑file refactoring – understands large codebases and performs consistent changes.

Low memory footprint via hierarchical compression.

Deep Git integration – auto‑generated commit messages, code review, PR management.

Easy model switching (Sonnet for everyday work, Opus for complex architecture, Haiku for quick exploration).

Drawbacks

Steep learning curve – many commands and shortcuts to master.

Terminal‑only interface – no GUI, which may deter developers unfamiliar with CLI tools.

Token cost can be high in heavy‑usage scenarios; requires conscious token budgeting.

Limited native Windows support; requires WSL or Git Bash.

Ideal Use Cases

Large‑scale project development – cross‑file refactoring, architecture analysis, complex bug hunting.

Automated workflows – CI/CD integration, code review, PR automation.

Multi‑task parallelism – background task handling for concurrent jobs.

Monorepo or micro‑service environments – --add-dir enables multi‑directory awareness.

Practical End‑to‑End Workflow (Spring Boot Auth Module Example)

# Navigate to project
cd my-spring-boot-project

# Start Claude Code
claude

# Initialise project knowledge base
/init

# Add project‑specific rules (e.g., coding style, security policies)
/memory

# Switch to Plan mode and ask for a design document
Shift+Tab   # Cycle to Plan mode
> Design a JWT authentication module and output a design document.

# Review the design, then switch to Auto‑Accept for implementation
Shift+Tab   # Cycle to Auto‑Accept mode
> Implement according to the design.

# Review changes and commit
/diff
/review
/commit

# Run compilation in background while continuing discussion
Ctrl+B   # Send compilation task to background

# When the conversation grows long, check token usage and compress
/context
/compact retain authentication design

# Or start a fresh task
/clear

Conclusion

Mastering roughly 30 % of the most frequently used commands (session management, Git automation, model switching, and skill handling) covers over 80 % of everyday development scenarios. Start with session commands ( /init, /compact, /clear, claude -c), add Git helpers ( /commit, /review), then explore model selection ( /model) and extensions ( /skills, /mcp). With these tools Claude Code becomes a true all‑round development partner.