The Claude Code starter kit
Start small. Build with care.
Three paths into responsible human–AI collaboration. Choose the one that meets you where you are.
Choose your starting point
Who are you today?
You're brand new to this. You're not sure what it is or what to do first. This guide walks you through your first 30 minutes - no experience needed, no jargon, just what to actually do.
You'll leave with
- A working CLAUDE.md file
- Your first real conversation
- The confidence to keep going
You've had a few conversations and you're ready to build something durable. Skills, memory, agents, governance. This is the playbook for turning Claude Code into an actual working system.
You'll leave with
- A complete CLAUDE.md setup
- Your first skill and agent
- A memory system that compounds
You've built the basics and you're thinking at OS scale. Subgroup design, routing governance, autonomous overnight patterns, approval gate architecture. This is for engineers who want to go all the way.
You'll leave with
- Subgroup and routing architecture
- Overnight session patterns
- Production safety frameworks
About this resource
Built from real experience. Shared freely.
This starter kit comes from months of building a real human-AI collaboration OS from scratch - making the mistakes, finding the patterns, encoding what worked. It's offered as open knowledge because the people who figure this out early have a real advantage, and that advantage shouldn't be hidden behind a paywall.
rafikiaos.com is a space for responsible human-agentic collaboration - practical resources, honest failures, and the patterns worth sharing. Everything here is free. Everything is real.
Keep exploring
