Claude Code Quickstart

A battle-tested starter kit for Claude Code. Clone it, customize it, and get productive in 60 seconds.

This is not a link collection or a config dump. It's an opinionated, progressive setup that teaches you the patterns that actually matter — from your first CLAUDE.md to automated guardrails and AI-powered code review.

Prerequisites

• Claude Code installed — pick one method:
  • curl -fsSL https://claude.ai/install.sh | bash (recommended)
  • brew install --cask claude-code (macOS)
• Active subscription — Claude Pro, Max, Team, or Enterprise — or an Anthropic Console account with API credits
• Git installed
• A terminal — macOS, Linux, or Windows (WSL recommended)

⚠️ NPM installation is deprecated. Use the native installer or Homebrew.

Full install docs: code.claude.com

Quick Start

# 1. Install Claude Code (if you haven't already)
curl -fsSL https://claude.ai/install.sh | bash

# 2. Clone and enter
git clone https://github.com/EduardPetraeus/claude-code-quickstart.git
cd claude-code-quickstart

# 3. Start Claude Code
claude

# 4. Verify — ask Claude:
#    "What conventions does this project follow?"

See QUICKSTART.md for detailed setup, verification steps, and troubleshooting.

Not a programmer? Start here → IMPLEMENT.md. Copy one prompt into Claude Code and it sets up everything for you.

How It Works

Claude Code reads your project configuration in layers — each layer adds structure:

CLAUDE.md                  → Project brain. Loaded every request. Conventions, rules, context.
.claude/rules/*.md         → Topic files. Git workflow, code style, security, testing.
.claude/hooks/*.sh         → Automation. Enforced on every tool call (lint, protect, scan).
.claude/agents/*.md        → Specialists. Code review, security audit, testing, exploration.

CLAUDE.md is the entry point — Claude reads it automatically. Rules add depth without bloating the main file. Hooks enforce guardrails you don't have to think about. Agents handle tasks that need focused expertise.

What's Inside

• CLAUDE.md template — Commented, ready to customize for your project
• Rules (8) — Git workflow, code style, session discipline, testing, Python conventions, web safety, session end protocol, security
• Hooks (10) — Auto-lint, branch protection, critical file protection, pre-push review gate, secret scanning, context monitoring, test reminders, session lifecycle, file suggestions
• Agents (9) — Code reviewer, security reviewer, explorer, quality gate, unit/integration/data/UAT/regression testers
• Commands (2) — Session handover (/project:handover) and reflection (/project:reflect)
• Prompts (8) — Battle-tested templates including 5 divergent thinking frameworks
• Guides (3) — Agentic engineering patterns, multi-AI adversarial review, parallel execution
• Tools (1) — Gemini-powered external code review CLI
• Docs (7) — Why each piece matters, with examples and a cheat sheet
• Exercises (4) — Hands-on exercises from beginner to advanced
• Examples (3) — CLAUDE.md templates for Python, TypeScript, and non-coders

Use It in Your Own Project

# 1. Copy the setup to your project
cp CLAUDE.md /path/to/your-project/
cp -r .claude/ /path/to/your-project/

# 2. Edit CLAUDE.md — fill in the <!-- CUSTOMIZE --> sections
#    (project name, stack, language preferences, project-specific rules)

# 3. Start Claude Code in your project
cd /path/to/your-project && claude

Create a CLAUDE.local.md for personal preferences (editor, language, local paths) — it's auto-gitignored and never committed.

Who This Is For

• Developers who want Claude Code to follow their conventions from day one
• Teams who want consistent AI behavior across projects
• Non-coders (HR professionals, analysts, writers, economists) who want a structured AI workflow — see IMPLEMENT.md
• Workshop participants who want a ready-to-go learning environment

Workshop

This repo powers a 2.5-hour hands-on workshop. See docs/workshop-guide.md for the full presenter guide with timed blocks, demo scripts, and exercises.

Structure

CLAUDE.md                        # The template — start here
QUICKSTART.md                    # Setup guide with verification
IMPLEMENT.md                     # Non-coder bootstrap — paste into Claude Code
.claude/
  settings.json                  # Full hooks + file suggestion config
  commands/                      # 2 custom slash commands (handover, reflect)
  rules/                         # 8 topic-specific rule files
  hooks/                         # 10 automation scripts
  agents/                        # 9 specialist agent definitions
prompts/                         # 8 prompt templates + divergent thinking
guides/                          # 3 deep pattern guides
tools/                           # External review CLI
docs/                            # 7 docs + cheat sheet
examples/                        # CLAUDE.md for Python, TypeScript, non-coders
exercises/                       # 4 hands-on exercises

Resources

• Claude Code Overview — official docs
• Claude Code Best Practices — Anthropic's recommended patterns
• Hooks Documentation — event-driven automation
• Settings Reference — permissions and configuration
• Cheat Sheet — keyboard shortcuts, commands, tips

Contributing

Issues and PRs are welcome. If you've built a useful rule, hook, or agent — share it.

License

MIT