Master Claude Code
The only comprehensive AI coding assistant guide you need
14,900+ lines • 88 templates • 227 quiz questions
Not a developer? Claude Cowork automates Excel, Word & PDF for knowledge workers.
Start With Interactive Onboarding
Claude asks 2-4 questions about your profile, then guides you through personalized content
claude "Fetch and follow the onboarding instructions from: https://raw.githubusercontent.com/FlorianBruniaux/claude-code-ultimate-guide/main/tools/onboarding-prompt.md"Choose your language
English or French
Select your goal
Learn, audit setup, or explore
Pick your level
Junior, Senior, or Power User
Get personalized path
Curated content for you
Quick Reference
Already know what you need?
| I need... | Go here | Time |
|---|---|---|
| Quick reference | Cheat Sheet | 2 min |
| Key principles | 6 Golden Rules | 1 min |
| Full documentation | Ultimate Guide | ~3h |
Starter CLAUDE.md Template (30 sec)
# Project: [NAME]
## Tech Stack
- Language: [e.g., TypeScript]
- Framework: [e.g., Next.js 14]
- Testing: [e.g., Vitest]
## Commands
- Build: `npm run build`
- Test: `npm test`
- Lint: `npm run lint`
## Rules
- Run tests before marking tasks complete
- Follow existing code patterns
- Keep commits atomic and conventionalSave as CLAUDE.md in your project root. Claude reads it automatically.
Why This Guide?
Claude Code transforms your terminal into an AI pair programming environment. Stop wasting time on trial and error.
Diagram by Mohamed Ali Ben Salem
Without the Guide
- Generic prompts that miss context
- Default config leaving performance on the table
- Debugging by trial and error
- Scattered tutorials, outdated in weeks
- No clear learning progression
With the Guide
- Battle-tested prompt patterns
- Optimized CLAUDE.md configurations
- Proven methodologies (TDD, SDD, BDD)
- 17,600+ lines of curated, maintained content
- Clear path from beginner to power user
Claude Code vs GitHub Copilot vs Cursor
How does this AI coding assistant compare to alternatives?
| Feature | Claude Code | GitHub Copilot | Cursor |
|---|---|---|---|
| Interface | Terminal CLI | IDE Extension | Full IDE (VS Code fork) |
| Agentic Mode | Multi-file, autonomous | Agent Mode (VS Code) | Composer mode |
| Context Window | 200K tokens (up to 1M) | 64K-128K tokens | 32K-200K tokens |
| Project Memory | CLAUDE.md files | ~ Repo embeddings | .cursorrules |
| Custom Agents | Agent SDK + Skills | ~ MCP configs | ~ MCP integrations |
| MCP Integration | Native client/server | Via Agent Mode | Docs integration |
| Shell Execution | Native terminal | Agent Mode | Integrated |
| Hooks System | Event-driven hooks | ||
| Best For | Terminal power users, automation | IDE-integrated coding | Visual IDE users |
Each tool has its strengths. Claude Code excels at autonomous multi-file operations and deep code generation from the terminal. This guide helps you master its unique capabilities.
Complementary AI Ecosystem
Claude Code excels at implementation. Chain it with the right tools for a complete workflow.
What's Inside
Everything you need to become proficient
Ultimate Guide
Comprehensive reference covering every aspect of Claude Code
Cheat Sheet
One-page printable reference for daily use
Popular80 Templates
Production-ready agents, commands, hooks, and skills
Interactive227-Question Quiz
Test your knowledge across all skill levels
Security & Privacy
Data handling, hardening guides, and best practices
61 Production Templates
Browse, preview, and copy with one click
Agents
Custom AI personas for code review, testing, security, DevOps
Commands
Slash commands for commits, PRs, tests, diagnostics
Hooks
Security checks, auto-formatting, injection detection
CI/CD
GitHub Actions for auto-review, security scan, triage
Config
Settings, Model Context Protocol (MCP) servers, gitignore patterns
Scripts
Audit, diagnostics, session search, health checks
Choose Your Path
Learning tracks adapted to your experience
Junior Developer
Foundation path for newcomers
- Read the Cheat Sheet
- Create your first CLAUDE.md
- Practice with simple tasks
- Learn diff review workflow
- Explore slash commands
- Take the beginner quiz
Senior Developer
Intermediate efficiency boost
- Skim Quick Start
- Set up custom hooks
- Configure MCP servers
- Create project-specific agents
- Master Plan Mode
- Build automation workflows
Power User
Comprehensive mastery path
- Read the full Ultimate Guide
- Build custom skills
- Create multi-agent systems
- Implement security hooks
- Design team workflows
- Contribute templates back
- Pass all quiz levels
Product Manager
Overview for non-technical roles
- Read Architecture overview
- Understand capabilities
- Review privacy guide
- Learn team integration
- Explore use cases
DevOps / SRE
Infrastructure and reliability focus
- Setup Claude for infrastructure
- Master the FIRE debugging framework
- Practice K8s troubleshooting
- Automate incident response
Product Designer
Design-to-code workflow with Figma MCP
- Learn image analysis basics
- Explore wireframing tools
- Setup Figma MCP
- Master design-to-code workflow β
- Print cheat sheet
6 Golden Rules
Memorize these for immediate improvement
Review diffs before accepting
Always check what Claude changes. Trust, but verify.
/compact at 70% context
Check with /status. Summarize before you hit the limit.
Be specific: WHAT + WHERE + HOW + VERIFY
Clear prompts get better results. Include success criteria.
Plan Mode first for complex tasks
Use Shift+Tab twice for risky or multi-step work.
Create CLAUDE.md for every project
Project context makes Claude significantly more effective.
Commit frequently after each task
Small atomic commits let you safely revert if something goes wrong.
The Prompt Formula
Structure your prompts for better results
Example
Add input validation to the login form.
WHERE: src/components/LoginForm.tsx
HOW: Use Zod schema, show inline errors
VERIFY: Empty email shows error messageLevel Up
Stop repeating yourself. Automate your patterns.
Automate repetition β Agents & Commands
Same review prompt 10x? Create an agent once, reuse forever. /commit, /pr, /review β your workflow, automated.
Extend knowledge β Skills
Teach Claude your stack conventions, TDD workflow, or security checklist. Reusable knowledge modules that persist across sessions.
Secure operations β Hooks
Block dangerous commands, scan for secrets, auto-format code. Event-driven scripts that run before/after Claude actions.
Combine tools β Ecosystem
Chain Claude Code with Perplexity, Gemini, NotebookLM for complete research-to-delivery workflows.
Community & Resources
Communaute francophone | Tu veux parler de dev et d'IA ? Challenger des idees ? Presenter des outils ? En decouvrir ?
Dev With AI rassemble pres de 1500 devs sur Slack, tous niveaux et toutes technos confondus β enthousiastes comme sceptiques. Meetups a Paris, Bordeaux et Lyon.
Contribute to the Guide
The guide repo is open-source. Found an error? Have a better template?
Contributing Guide →
Claude Code Releases
Latest versions and breaking changes from Anthropic
Showing 6 of 42 releases
- New: CLAUDE_CODE_ENABLE_TASKS env var (toggle task systems)
- New: Argument shorthand $0, $1 in custom commands
- [VSCode] Session forking/rewind for all users
- Fixed: Multiple session resuming issues with git worktrees
- β Customizable keyboard shortcuts with /keybindings
- Per-context keybindings, chord sequences
- Fix: Crashes on processors without AVX instruction support
- β New task management system with dependency tracking
- [VSCode] Native plugin management support
- [VSCode] OAuth users can browse/resume remote sessions
- Fixed: OOM crashes with heavy subagent usage
- β οΈ Deprecation notice for npm installations
- UI rendering performance improved with React Compiler
- Fixed: MCP stdio server timeout causing UI freezes
- History-based autocomplete in bash mode (! + Tab)
- Search in installed plugins list
- Git commit SHA pinning for plugins
- Multiple fixes: context window, memory, autocomplete
- Bug fix: Message rendering
- Fix: Excessive MCP connection requests for HTTP/SSE transports
- New
Setuphook event (--init, --maintenance flags) - Keyboard shortcut 'c' to copy OAuth URL
- File suggestions show as removable attachments
auto:Nsyntax for MCP tool search thresholdplansDirectorysetting for custom plan locations- Session URL attribution to commits/PRs
showTurnDurationsetting- MCP tool search auto mode enabled by default
- Inline agent response in task notifications
- Security fix: Wildcard permission rules now properly handle compound commands (cd && rm) requiring multiple rules
- OAuth/API Console: console.anthropic.com β platform.claude.com (credential migration required)
- Search in
/configcommand - Date range filtering in
/stats - Auto-discovery of nested
.claude/skills
- Security fix: Prevented shell line continuation from bypassing permission checks
- URL validation for remote sessions
- Fixed: Agent tools with binary outputs now handle properly
- Agent tool output encoding improvements
- Fixed: Remote sessions error handling
/infocommand displays MCP server list- MCP tool discovery improvements
- Enhanced error messages for MCP server failures
- Fixed: Bash tool permission edge cases
- Security fix: Command injection vulnerability in bash processing
- Improved agent task output formatting
- Fixed: Permission prompt edge cases
- Skill hot-reload without restart
- Vim motions support in editor
/plancommand for structured planning
- OAuth/API Console URL changed: console.anthropic.com β platform.claude.com
- Agent task UI improvements
- Fixed: Skill command parsing
- Removed # shortcut for quick memory (use /memory command instead)
- MCP server status indicators
- Enhanced bash command history
- Agent tool output formatting
- Fixed: Skill discovery edge cases
- Improved error messages for permission denials
- Fixed: MCP tool schema validation
- Enhanced agent task progress indicators
- Fixed: Bash tool timeout handling
- MCP server health checks
- Fixed: Skill argument parsing
- Improved agent task error recovery
- Fixed: Permission prompt display
- Enhanced bash autocomplete suggestions
- Fixed: MCP tool discovery
- Agent tool output formatting improvements
- Fixed: Skill hot-reload edge cases
- Improved error messages for MCP server failures
- Fixed: Permission prompt edge cases
- Enhanced agent task progress indicators
- Fixed: Bash tool timeout handling
- MCP server health checks
- Fixed: Skill argument parsing
- Improved agent task error recovery
- Fixed: Permission prompt display
- Enhanced bash autocomplete suggestions
- Fixed: MCP tool discovery
- Agent tool output formatting improvements
- Fixed: Skill hot-reload edge cases
- Improved error messages for MCP server failures
- Fixed: Permission prompt edge cases
- Enhanced agent task progress indicators
- Fixed: Bash tool timeout handling
- MCP server health checks
- Fixed: Skill argument parsing
- Improved agent task error recovery
- Fixed: Permission prompt display
- Enhanced bash autocomplete suggestions
- Fixed: MCP tool discovery
- Agent tool output formatting improvements
- Fixed: Skill hot-reload edge cases
- Improved error messages for MCP server failures
- Fixed: Permission prompt edge cases
- Enhanced agent task progress indicators
- Fixed: Bash tool timeout handling
- MCP server health checks
- Fixed: Skill argument parsing
⚠ Recent Breaking Changes
- Syntax Indexed argument syntax changed: $ARGUMENTS.0 → $ARGUMENTS[0] (v2.1.19)
- Install npm installations deprecated - use native installer (v2.1.15)
- Security Command injection fix in bash processing (v2.1.2)
- Security Wildcard permission rules compound commands fix (v2.1.7)
- Security Shell line continuation permission bypass fix (v2.1.6)
- OAuth OAuth/API Console: console.anthropic.com → platform.claude.com (v2.1.0, v2.1.7)
- Removed Removed # shortcut for quick memory (v2.0.70)
Not a Developer?
Claude Cowork is designed for knowledge workers who need AI-powered file automation without writing code.
Hitting Rate Limits or High Costs?
cc-copilot-bridge routes Claude Code through GitHub Copilot Pro+ for flat-rate access.
FAQ
What is Claude Code and how does it work?
Claude Code is Anthropic's official CLI (command-line interface) that brings Claude AI directly into your terminal. It acts as an AI coding assistant that can read your codebase, write code, run commands, and make multi-file changes autonomously. Think of it as AI pair programming in your terminal.
How do I install Claude Code?
Install globally via npm: npm install -g @anthropic-ai/claude-code. Requires Node.js 18+. Then run claude in any project directory to start. See the cheatsheet for a quick reference.
Is Claude Code free to use?
Claude Code requires an Anthropic API key with a paid plan. The tool itself is free, but you pay for API usage. Sonnet is the default model (cost-effective for most tasks). Opus is available for complex architecture work at higher cost.
How can I reduce costs or avoid rate limits?
For heavy usage, cc-copilot-bridge routes Claude Code through GitHub Copilot Pro+ ($10/month flat) instead of per-token billing. This can save 99%+ on costs and bypass Anthropic rate limits. It also supports offline mode with Ollama for air-gapped environments.
Is Claude Code better than GitHub Copilot?
Both now have agentic capabilities. Claude Code excels at terminal workflows with native MCP, CLAUDE.md memory, Agent SDK, and hooks. Copilot integrates well with VS Code via Agent Mode. Many developers use both. See our comparison table for details.
What is a CLAUDE.md file?
CLAUDE.md is a project memory file that gives Claude persistent context about your project. Place it in your project root with your tech stack, coding conventions, and commands. Claude reads it automatically at session start. See our templates.
What's the difference between Claude Code and Cowork?
Claude Code is a terminal-based CLI for developers (code, scripts, git). Cowork is a Desktop app for knowledge workers (Excel, Word, PDF). This guide covers Claude Code. See the Cowork section in the guide for details.
What are MCP servers in Claude Code?
MCP (Model Context Protocol) servers extend Claude's capabilities by connecting it to external tools and data sources. Examples include database access, browser automation, and documentation lookup. Configure them in ~/.claude/mcp.json.
What programming languages does Claude Code support?
Claude Code works with any programming language. It's particularly strong with TypeScript, JavaScript, Python, Go, Rust, and Ruby. It can read, write, and refactor code in any language you use.
Is this guide official?
No, this is a community resource. It reflects personal experience after months of daily Claude Code usage, not official Anthropic documentation.
What version of Claude Code is covered?
oThe guide is continuously updated. Current version: 3.11.6. Check the CHANGELOG for recent updates.
Is my code sent to Anthropic?
Yes, Claude Code sends prompts and file contents to Anthropic servers. Default retention is 5 years (training enabled). You can opt-out to 30 days or use Enterprise for 0 retention. See the privacy guide.
Who is behind this guide?
I'm Florian BRUNIAUX, a tech professional with 11+ years of experience across roles including CTO, VP Engineering, Engineering Manager, and Founding Engineer. I've built and scaled teams from scratch (up to 30 people), worked on B2B/B2C SaaS products, and I'm deeply passionate about AI-assisted development. I created this guide as an open-source contribution to help the community master Claude Codeβthe same way books like Clean Architecture, The Phoenix Project, and The Manager's Path helped me grow. Learning in public and giving back is what drives me.
Which AI tools complement Claude Code?
Claude Code excels at implementation but works best in an ecosystem: Perplexity for deep research with verified sources, Gemini for visual understanding (design → code), NotebookLM for audio synthesis and doc review, v0/Bolt for rapid UI prototyping, Cursor/Copilot for inline autocomplete, and Wispr Flow for voice-to-text (4x faster than typing). Chain them: Research (Perplexity) → Implement (Claude Code) → Deliver (Kimi for slides). See the full ecosystem guide.
How do I use voice-to-text with Claude Code?
Use tools like Wispr Flow or Superwhisper to dictate prompts ~4x faster than typing (150 WPM vs 40 WPM). These tools transcribe your speech directly into the terminal. The AI Ecosystem guide covers setup and integration patterns.
How much should I trust AI-generated code?
Trust varies by risk level: auto-accept boilerplate and repetitive patterns, review business logic and non-trivial changes, manually verify security-critical code. Research shows even experienced developers miss bugs in AI code. See the Trust Calibration guide for research-backed verification strategies.
Can I contribute to the guide?
Yes! The guide repository is open-source and welcomes contributions. Found an error? Have a better template? Check the contributing guide on GitHub. Note: this website is maintained separately by the author.