The Ultimate Claude Code Cheatsheet

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

# Verify installation
claude --version

# ── Authentication Methods ──

# Method 1: Anthropic API Key (set environment variable)
export ANTHROPIC_API_KEY="sk-ant-..."

# Method 2: Claude Pro/Max subscription
claude auth login
# Opens browser → authenticate with Anthropic account

# Method 3: Amazon Bedrock (for AWS users)
export ANTHROPIC_BEDROCK_BASE_URL="https://..."
export AWS_REGION="us-east-1"
export AWS_ACCESS_KEY_ID="..."
export AWS_SECRET_ACCESS_KEY="..."

# Method 4: Google Vertex AI
export ANTHROPIC_VERTEX_PROJECT="your-project"
export ANTHROPIC_VERTEX_REGION="us-east5"

# ── Start Claude Code ──

# Interactive REPL
claude

# With a specific model
claude --model claude-sonnet-4-20250514

# Non-interactive (piped / one-shot)
echo "Fix the bug in app.ts" | claude
claude -p "Explain this codebase in 3 sentences"

# Resume a previous conversation
claude --resume

# Continue the last conversation
claude --continue

# ── CI/CD Integration ──
claude -p "Run tests and report failures" --max-turns 10 --allowedTools "Bash"

# ── Batch Processing ──
for file in src/**/*.test.ts; do
  claude -p "Fix failing test in $file" --max-turns 5
done

# ── JSON Output for Scripts ──
claude -p "List all API endpoints" -o json | jq '.response'

# ── Git Workflow ──
claude -p "Review this PR: $(git diff main)" --model claude-opus-4-20250514

# Project: My Next.js App

## Tech Stack
- Framework: Next.js 15 with App Router
- Language: TypeScript (strict mode)
- Styling: Tailwind CSS 4 + shadcn/ui
- Database: Prisma ORM with SQLite
- Package Manager: pnpm

## Coding Rules
- Use 'use client' for all interactive components
- Prefer shadcn/ui components over custom implementations
- Use Zustand for client state, TanStack Query for server state
- All API routes use Next.js Route Handlers (not Server Actions)
- No blue/indigo colors unless explicitly requested

## Commands
- Install deps: pnpm install
- Run dev: pnpm dev
- Lint: pnpm lint
- Type check: pnpm tsc --noEmit
- Run tests: pnpm test

## Code Style
- Use early returns in functions
- Prefer const over let
- All components use PascalCase
- Utility functions in src/lib/
- Types/interfaces in src/types/

## Git
- Commit messages follow Conventional Commits
- Always run lint before committing

# Code Review Skill

## Trigger
Use when the user asks for a code review or PR review.

## Process
1. Read all changed files in the PR/diff
2. Analyze each file for the following categories:

### Security
- SQL injection, XSS, CSRF vulnerabilities
- Hardcoded secrets, API keys, passwords
- Improper input validation
- Insecure dependency versions

### Performance
- N+1 queries, unnecessary re-renders
- Missing indexes, unbounded loops
- Memory leaks, large bundle imports

### Code Quality
- DRY violations, dead code
- Naming conventions, type safety
- Error handling completeness
- Test coverage gaps

## Output Format
Provide findings in a table:
| Severity | File | Line | Issue | Suggestion |

# Deployment Skill

## Trigger
Use when the user asks to deploy the project.

## Steps
1. Run lint: pnpm lint
2. Run type check: pnpm tsc --noEmit
3. Run tests: pnpm test
4. Build: pnpm build
5. Check for uncommitted changes: git status
6. Create release tag: git tag v${VERSION}
7. Deploy to Vercel: vercel --prod

## Rollback
If any step fails:
- Stop immediately
- Report the failure
- Suggest rollback command: vercel rollback

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Write",
        "command": "echo 'About to write a file...'"
      },
      {
        "matcher": "Bash(git commit *)",
        "command": "pnpm lint && pnpm test",
        "if": "Bash(git commit *)"
      },
      {
        "matcher": "Bash(npm publish *)",
        "command": "echo 'BLOCKED: Use pnpm publish instead'",
        "exit_code": 1
      }
    ],
    "PostToolUse": [
      {
        "matcher": "Write",
        "command": "pnpm prettier --write $FILE"
      },
      {
        "matcher": "Write(src/**/*.ts)",
        "command": "pnpm tsc --noEmit"
      }
    ],
    "Notification": [
      {
        "matcher": "",
        "command": "osascript -e 'display notification "$MESSAGE"'"
      }
    ]
  }
}

{
  "permissions": {
    "allow": [
      "Read",
      "Write(src/**)",
      "Write(src/components/**)",
      "Edit(src/**)",
      "Bash(git *)",
      "Bash(pnpm *)",
      "Bash(node *)",
      "mcp__github__*"
    ],
    "deny": [
      "Bash(rm -rf *)",
      "Bash(sudo *)",
      "Write(.env*)",
      "Write(secrets/**)",
      "Bash(npm publish *)",
      "Bash(curl * | bash)"
    ],
    "ask": [
      "Write",
      "Bash(*)",
      "mcp__*"
    ]
  }
}

{
  "permissions": {
    "allow": [
      "Bash(claude *)"
    ]
  }
}

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/mcp-github"],
      "env": {
        "GITHUB_TOKEN": "${GITHUB_TOKEN}"
      }
    },
    "postgres": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/mcp-postgres", "postgresql://..."]
    },
    "puppeteer": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/mcp-puppeteer"]
    },
    "remote-server": {
      "type": "sse",
      "url": "https://my-mcp-server.example.com/sse"
    }
  }
}

# ── Auto Mode Examples ──

# Migrate all .js files to .ts
claude --auto "Migrate all files in src/ from JavaScript to TypeScript"

# Generate tests for all components
claude --auto "Write unit tests for every component in src/components/"

# Refactor API to new pattern
claude --auto "Refactor all API routes to use the new error handling pattern"

# With limits
claude --auto --max-turns 50 "Add TypeScript types to all utility functions"

# Computer Use: E2E testing
claude "Open localhost:3000 in the browser, test the login flow, report any issues"

{
  "scheduledTasks": {
    "daily-review": {
      "schedule": "0 9 * * *",
      "prompt": "Review all commits from the last 24 hours for issues",
      "model": "claude-sonnet-4-20250514",
      "max-turns": 20
    },
    "weekly-deps": {
      "schedule": "0 10 * * 1",
      "prompt": "Check for outdated dependencies and suggest updates",
      "allowedTools": ["Read", "Bash(pnpm *)"]
    }
  }
}

{
  "model": "claude-sonnet-4-20250514",
  "permissions": {
    "allow": ["Read", "Bash(git *)"],
    "deny": ["Bash(rm -rf *)"],
    "ask": ["Write", "Bash(*)"]
  },
  "hooks": {
    "PreToolUse": [
      { "matcher": "Write", "command": "pnpm lint" }
    ]
  },
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/mcp-github"]
    }
  }
}

# ── Diagnostic Commands ──

# Check Claude Code version
claude --version

# Run diagnostics
claude /doctor

# Check auth status
claude auth status

# Clear all cached data
rm -rf ~/.claude/cache
rm -rf ~/.claude/conversations

# Reset permissions for current project
rm .claude/claude.json

# Test MCP server connectivity
claude -p "List available MCP tools"

# Check Node.js version (requires 18+)
node --version

# Reinstall Claude Code
npm install -g @anthropic-ai/claude-code@latest

# Verbose logging for debugging
claude --verbose 2>&1 | tee claude-debug.log