본문으로 건너뛰기

Claude Code: Quick Reference Guide

Installation & Setup​

# Install
npm install -g @anthropic-ai/claude-code

# Start
cd your-project
claude

Core Commands​

# Interactive mode
claude

# Single command
claude -p "your prompt here"

# Continue previous conversation  
claude --continue

# Resume from saved state
claude --resume

Slash Commands​

  • /help - Get help and documentation
  • /mcp - View MCP server status and tools
  • /permissions - Manage tool permissions
  • /export - Export conversation history
  • /status - Show system status
  • /config - View configuration
  • /doctor - Diagnose system issues
  • /bug - Report bugs

Configuration Files​

~/.claude/settings.json          # User settings
.claude/settings.json            # Project settings  
.claude/commands/*.md            # Custom commands
.mcp.json                        # MCP server config

Essential Settings​

{
  "model": "claude-3-5-sonnet-20241022",
  "provider": "anthropic", 
  "maxTokens": 4096,
  "temperature": 0.7
}

Built-in Tools​

  • BashTool - Execute shell commands
  • SearchTool - Search codebase with regex
  • WebTool - Fetch web content
  • TaskTool - Manage TODO lists
  • ReadTool - Read file contents
  • EditTool - Modify files
  • Git Integration - Version control

Common Workflows​

Feature Development​

claude -p "Add a user authentication system with login/logout functionality"

Debugging​

claude -p "Fix the memory leak in the UserService class"

Code Review​

claude -p "Review this pull request and suggest improvements"

Refactoring​

claude -p "Refactor the payment processing module to use the Strategy pattern"

MCP Integration​

{
  "mcpServers": {
    "github": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "github-mcp-server"],
      "env": {"GITHUB_TOKEN": "$GITHUB_TOKEN"}
    }
  }
}

Hook System​

# Available hooks
- UserPromptSubmit
- PreToolUse  
- PreCompact
- Stop
- SubagentStop

Environment Variables​

ANTHROPIC_API_KEY=your-key
CLAUDE_CONFIG_DIR=~/.config/claude-code
AWS_BEARER_TOKEN_BEDROCK=token
CLAUDE_MODEL=claude-3-5-sonnet-20241022

Troubleshooting​

Permission Issues​

claude /permissions  # Check current permissions

MCP Server Problems​

claude /mcp          # View MCP server status

Configuration Problems​

claude /doctor       # Run diagnostics

Token Limits​

  • Use focused prompts
  • Leverage /memory for context
  • Choose appropriate model for task

Best Practices​

Prompting​

  • Be specific and actionable
  • Include context and constraints
  • Break complex tasks into steps

Security​

  • Review generated code
  • Use git for change tracking
  • Configure permission restrictions
  • Store API keys securely

Performance​

  • Start with small tasks
  • Use appropriate models
  • Monitor token usage
  • Leverage caching when possible

Common Patterns​

Multi-step Development​

claude -p "Create a React component for user profiles, add unit tests, and update the main app to use it"

Error Investigation​

claude -p "The application is throwing 'Cannot read property of undefined' errors. Investigate and fix the root cause."

Documentation​

claude -p "Generate comprehensive documentation for the API endpoints in src/routes/"

Code Quality​

claude -p "Run linting, fix all warnings, and ensure the code follows project conventions"

Integration Examples​

GitHub Actions​

- uses: anthropics/claude-code-action@v1
  with:
    prompt: "Review this PR and add appropriate labels"  
    mcp_config: .mcp.json
    allowed_tools: ["mcp__github__*", "Bash"]

VS Code Integration​

{
  "claude.autoComplete": true,
  "claude.contextAware": true,
  "claude.model": "claude-3-5-sonnet-20241022"
}

Model Selection Guide​

Sonnet 3.5 (Default)​

  • General coding tasks
  • Code review and debugging
  • Documentation generation

Opus 4​

  • Complex architectural decisions
  • Advanced refactoring
  • Research and analysis

Sonnet 4​

  • Latest features
  • Enhanced reasoning
  • Improved codebase understanding