Complete Claude Code Workflow & Best Practices Cheat Sheet

This comprehensive reference guide consolidates all the essential patterns, commands, and workflows from our entire conversation about Claude Code for maximum productivity and effectiveness.

Table of Contents

1. Quick Start Setup
2. The Three Essential Directories Pattern
3. Core Commands & Modes
4. Thinking Modes & Keywords
5. Subagents & Parallel Processing
6. Custom Slash Commands
7. MCP Server Integration
8. Apple Development Workflow
9. Best Practice Workflows
10. Testing & Quality Assurance
11. Quick Reference Tables
12. Troubleshooting & Tips

Quick Start Setup

Installation & Authentication

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

# Launch Claude Code
claude

# Update to latest version
claude update

Essential First Steps

# 1. Create project memory file
touch CLAUDE.md

# 2. Initialize custom commands directory
mkdir -p .claude/commands

# 3. Create context prime command
echo "Read project files and run 'tree -I 'node_modules|.git|build|DerivedData'' to understand structure" > .claude/commands/context-prime.md

The Three Essential Directories Pattern

Context is King Philosophy

The three directories solve the fundamental problem: your AI coding tool must have the right information at the right time to be effective[1][2][3].

1. ai-docs/ - Your AI Agent's Persistent Memory

Purpose: Think of this as your AI coding tool's persistent knowledge base[3].

What Goes Inside:

• Third-party API documentation
• Integration details and patterns
• Custom conventions specific to your codebase
• Implementation notes and gotchas
• Apple frameworks documentation (SwiftUI, UIKit, etc.)

Example Structure:

ai-docs/
├── apple-frameworks-best-practices.md
├── third-party-apis.md
├── custom-patterns.md
└── troubleshooting-guide.md

2. specs/ - Your AI Coding Plans

Purpose: This is the most important folder in your entire codebase. Where you write great plans and scale up your compute[3].

Key Principle: The plan IS the prompt. Great planning equals great prompting[3].

What Goes Inside:

• Feature specifications (PRDs/Product Requirements Documents)
• Architecture plans
• UI/UX implementation specs
• Testing strategies
• Cross-platform integration requirements

Example Structure:

specs/
├── feature-user-authentication.md
├── cross-platform-sync.md
├── ui-component-library.md
└── testing-strategy.md

3. .claude/ - Your Reusable Prompts

Purpose: Store reusable, runnable prompts that save time and ensure consistency[2][3].

Essential Structure:

.claude/
└── commands/
    ├── context-prime.md          # Most important command
    ├── session-start.md
    ├── deploy-check.md
    ├── cross-platform-review.md
    └── testflight-prep.md

Core Commands & Modes

Interactive Mode Commands

Special Key Combinations

Print Mode (SDK Usage)

# Run single prompt and exit
claude -p "Write a function to calculate Fibonacci numbers"

# Process piped content
cat logs.txt | claude -p "analyze these logs"

# Output in JSON format
claude -p "generate hello world" --output-format json

# Resume specific session
claude -r "session-id" "continue this task"

Thinking Modes & Keywords

Thinking Budget Keywords[2][6]

Usage Example:

> ultrathink: Design a cross-platform authentication system for iOS, macOS, and watchOS with offline sync capabilities

When to Use Each Level

• think: Standard debugging, feature planning
• think hard: Architecture decisions, complex integrations
• think harder: Performance optimization, security reviews
• ultrathink: System design, complex algorithms, critical debugging[6]

Subagents & Parallel Processing

Core Concept

Subagents enable parallel processing, context management, and specialized task delegation within a single workflow[7][4].

Key Benefits

1. Parallel Processing: Multiple aspects investigated simultaneously
2. Context Isolation: Each subagent has its own context window
3. Specialization: Assign expert roles (security, accessibility, etc.)
4. Synthesis: Main agent compiles results into coherent solution

Verification Workflow Stages

Practical Example

> Implement user authentication. Spawn 3 subagents:
> 1. Security expert to review vulnerabilities
> 2. UX expert to validate user flow
> 3. Testing expert to verify edge cases
> Consolidate findings before proceeding.

Custom Slash Commands

Setup & Structure

# Personal commands (available across all projects)
~/.claude/commands/

# Project-specific commands
.claude/commands/

Essential Commands to Create

1. Context Prime (Most Important)

# .claude/commands/context-prime.md
Read the following files to understand this project:
- README.md
- Package.json/Package.swift
- Main architecture files
- Any existing integration notes

Then run `tree -I 'node_modules|.git|build|DerivedData'` to understand project structure.

2. Session Management

# .claude/commands/session-start.md
Initialize session and prepare work environment:
1. Check current status with git status
2. Get latest changes with git pull
3. Review previous work
4. Present today's work plan

3. Apple Development Specific

# .claude/commands/cross-platform-check.md
Review this feature for compatibility across:
- iOS (iPhone/iPad)
- macOS
- watchOS
- CarPlay (if applicable)

Check for:
- Shared business logic opportunities
- Platform-specific UI adaptations
- Data synchronization requirements
- Testing strategies for each platform

Command Templates with Arguments

# .claude/commands/add-tests.md
Generate comprehensive tests for $ARGUMENTS:
- Unit tests for core functionality
- Edge case testing
- Error handling validation
- Integration test scenarios

Usage: /project:add-tests UserAuthentication.swift

MCP Server Integration

What is MCP?

Model Context Protocol (MCP) connects Claude Code to external tools and data sources[8][9].

Setup Commands

# Add MCP stdio server
claude mcp add my-server -e API_KEY=123 -- /path/to/server arg1 arg2

# Add MCP SSE server
claude mcp add --transport sse api-server https://api.example.com/mcp --header "X-API-Key: your-key"

# Add MCP HTTP server
claude mcp add --transport http secure-server https://api.example.com/mcp --header "Authorization: Bearer token"

# List configured servers
claude mcp list

# Remove server
claude mcp remove my-server

Scope Options

• local: Available only to you in current project
• project: Shared with team via .mcp.json file
• user: Available across all your projects

Popular MCP Servers

• GitHub: PR review, issue management
• Sentry: Error tracking and debugging
• DigitalOcean: App deployment and management[9]
• Brave Search: Web search capabilities
• Fetch: HTTP request handling

Apple Development Workflow

Project Structure for Apple Development

your-ios-project/
├── ai-docs/
│   ├── swiftui-patterns.md
│   ├── apple-guidelines.md
│   ├── cross-platform-strategies.md
│   └── testflight-process.md
├── specs/
│   ├── ios-app-architecture.md
│   ├── watchos-companion.md
│   ├── shortcuts-integration.md
│   └── carplay-features.md
└── .claude/
    └── commands/
        ├── context-prime.md
        ├── cross-platform-review.md
        ├── testflight-prep.md
        └── shortcuts-integration.md

Apple-Specific Commands

Shortcuts Integration

# .claude/commands/shortcuts-integration.md
Analyze this feature for Shortcuts app integration:
- Identify functions suitable for Shortcuts exposure
- Create Intent definitions
- Design Siri phrase patterns
- Plan voice interaction flows
- Consider MCP server integration for AI systems

TestFlight Preparation

# .claude/commands/testflight-prep.md
Prepare this build for TestFlight:
- Review app metadata and descriptions
- Check privacy permissions and usage descriptions
- Validate testing instructions
- Ensure crash reporting is configured
- Generate testing scenarios checklist

Cross-Platform Development

# .claude/commands/cross-platform-architecture.md
Design feature architecture for Apple ecosystem:
- Shared business logic (Core Data, CloudKit)
- Platform-specific UI adaptations
- Data synchronization strategies
- Handoff and Continuity features
- Universal app considerations

iOS Development with Claude Code

# Start new iOS project
cd YourApp
claude

# Initialize project
/project:context-prime

# Enter planning mode
# Shift + Tab
> ultrathink: Design a SwiftUI app with watchOS companion that syncs data via CloudKit

# Execute approved plan
# Shift + Tab (again to exit plan mode)

Best Practice Workflows

1. Explore → Plan → Implement → Verify → Integrate

Explore Phase

> Read relevant files first before coding
> Use subagents for preliminary research
> /project:context-prime to establish baseline understanding

Plan Phase

> Shift + Tab (enter Plan Mode)
> ultrathink: [complex problem description]
> Create detailed specification in specs/ directory
> Do not code until plan is approved

Implement Phase

> Use auto-accept mode for approved plans
> Reference files explicitly (@path/to/file)
> Iterate in short, focused messages

Verify Phase

> Spawn specialized subagents for validation
> Security review, performance check, accessibility audit
> Cross-platform compatibility verification

Integrate Phase

> Consolidate insights into CLAUDE.md
> Update documentation
> Use /clear between major tasks

2. Test-Driven Development (TDD) Pattern[10]

> Write failing tests first
> Generate minimal implementation
> Refactor and optimize
> Verify with subagents

3. Plan Drafting Technique[3]

> Have Claude create the first draft of your plan
> Iterate and refine the plan
> Execute only after approval
> Document learnings for future reference

Testing & Quality Assurance

Testing Strategies with Claude Code

1. Comprehensive Test Generation

> Generate tests for EdgeCaseHandler.swift covering:
> - Normal operation scenarios
> - Edge cases and boundary conditions
> - Error handling and recovery
> - Performance under load
> - Integration with other components

2. Test-Driven Development Workflow[10][11]

> Write failing tests based on requirements
> Implement minimal code to pass tests
> Refactor with confidence
> Use subagents to verify test quality

3. Quality Assurance Checklist

• Unit tests for core functionality
• Integration tests for component interaction
• UI tests for user workflows
• Performance tests for critical paths
• Security tests for vulnerabilities
• Accessibility tests for compliance

Apple-Specific Testing

# .claude/commands/ios-testing-strategy.md
Create comprehensive testing strategy for iOS:
- XCTest unit tests for business logic
- XCUITest for user interface testing
- Performance testing with XCTMetric
- TestFlight beta testing plan
- App Store review preparation

Quick Reference Tables

Essential File Structure

Command Categories

Thinking Budget Guide

Troubleshooting & Tips

Common Issues & Solutions

1. Context Overflow

• Problem: Running out of context window
• Solution: Use /clear between tasks, break into smaller chunks

2. Slow Performance

• Problem: Claude Code responding slowly
• Solution: Reduce context size, use more focused prompts

3. Inconsistent Results

• Problem: Variable code quality
• Solution: Use plan mode, be more specific in prompts

4. Apple Development Issues

• Problem: Cross-platform compatibility
• Solution: Use cross-platform review commands, test on all target devices

Optimization Tips

1. Token Management

• Use custom commands to reduce repetitive context loading[12]
• Keep CLAUDE.md concise and focused
• Clear context regularly with /clear

2. Workflow Efficiency

• Start every session with /project:context-prime
• Use Plan Mode for complex features
• Leverage subagents for parallel validation

3. Quality Assurance

• Always verify plans before implementation
• Use specialized subagents for reviews
• Document learnings in ai-docs/

4. Team Collaboration

• Share .claude/commands/ across team
• Standardize CLAUDE.md format
• Use project-scoped MCP servers

Performance Monitoring

# Check token usage
claude --cost

# Monitor session performance
claude --verbose

# Review session history
claude --resume <session-id>

Advanced Patterns

1. Multi-Instance Parallel Development[4]

# Use git worktrees for parallel development
git worktree add ../feature-branch feature-branch
cd ../feature-branch
claude

2. Headless Automation[2]

# CI/CD integration
claude -p "run tests and generate report" --output-format json

# Batch processing
for file in *.py; do
  claude -p "review and optimize $file" < "$file"
done

3. Visual Development[4]

> Drag and drop screenshots into terminal
> Claude analyzes visual mockups
> Generates code matching designs

This comprehensive cheat sheet provides everything you need to master Claude Code for professional development workflows, with special emphasis on Apple ecosystem development patterns. Keep this reference handy and customize the patterns to fit your specific needs and projects[1][13][14][7][4][2][15][16][8][9][17][5][18][12][6][3][19][10][20][11].

Sources
[1] CLI reference - Anthropic API https://docs.anthropic.com/en/docs/claude-code/cli-reference
[2] Claude Code: Best practices for agentic coding - Anthropic https://www.anthropic.com/engineering/claude-code-best-practices
[3] Agentic Claude Code: 3 Codebase Folders for TOP 1% AI Coding https://www.youtube.com/watch?v=hGg3nWp7afg
[4] How I use Claude Code | Christian B. B. Houmann https://bagerbach.com/blog/how-i-use-claude-code/
[5] Yup, Claude Code Plan Mode is here: The Senior Engineer's Workflow https://www.youtube.com/watch?v=7LWl3EbcFTc
[6] The Secret Prompt To Unlock Claude Code GOD MODE - YouTube https://www.youtube.com/watch?v=ntc5O2x1IFE
[7] Anthropic Releases Best Practices Guide for Claude Code ... - AIbase https://www.aibase.com/news/www.aibase.com/news/17387
[8] Model Context Protocol (MCP) - Anthropic API https://docs.anthropic.com/en/docs/claude-code/mcp
[9] Setting Up the DigitalOcean MCP Server in Claude Code https://www.digitalocean.com/community/tutorials/claude-code-mcp-server
[10] Claude Code & Test-Driven Development: A Comprehensive Guide https://talent500.com/blog/claude-code-test-driven-development-guide/
[11] Mastering Claude Code: Essential Tips for Maximum Productivity https://www.tembo.io/blog/mastering-claude-code-tips
[12] Boost Your Development Efficiency with Claude Code Custom ... https://gizin.co.jp/en/tips/claude-code-custom-commands
[13] Claude Code overview - Anthropic API https://docs.anthropic.com/en/docs/claude-code/overview
[14] Claude Code SDK - Anthropic API https://docs.anthropic.com/en/docs/claude-code/sdk
[15] Common workflows - Anthropic API https://docs.anthropic.com/en/docs/claude-code/common-workflows
[16] Claude Code: A Complete Setup Guide and Honest Evaluation https://www.helicone.ai/blog/evaluating-claude-code
[17] Claude Code MCP Server MCP server for AI agents - Playbooks https://playbooks.com/mcp/claude-code
[18] Claude Code + MCP Setup: Late Night Tutorial ... - YouTube https://www.youtube.com/watch?v=oM2dXJnD80c
[19] How to Build an iOS App With Claude AI: A Step-by-Step Guide https://slashdev.io/-how-to-build-an-ios-app-with-claude-ai-a-step-by-step-guide
[20] How To Use Claude Code [Make Your First iOS App ... - YouTube https://www.youtube.com/watch?v=FLFfVxUUmLE
[21] Set up Claude Code - Anthropic API https://docs.anthropic.com/en/docs/claude-code/setup
[22] GitHub - anthropics/claude-code https://github.com/anthropics/claude-code
[23] How to get MCP servers running on Claude Code - Reddit https://www.reddit.com/r/ClaudeAI/comments/1iztm9b/how_to_get_mcp_servers_running_on_claude_code/
[24] Configuring MCP Tools in Claude Code - The Better Way https://scottspence.com/posts/configuring-mcp-tools-in-claude-code
[25] A Prompt for Smoother Claude Code Onboarding - Apidog https://apidog.com/blog/claude-code-onboarding-prompt
[26] The Claude Code Complete Guide: Learn Vibe-Coding & Agentic AI https://natesnewsletter.substack.com/p/the-claude-code-complete-guide-learn
[27] Best way to use Claude Projects for coding. One big file with ... - Reddit https://www.reddit.com/r/ClaudeAI/comments/1efy6yf/best_way_to_use_claude_projects_for_coding_one/