AI OS Setup Guide
Tools · 8 min
An AI OS turns Claude Code from a one-shot tool into a persistent operating system for your business. You need only Claude Code and Python 3.10+. Setup takes about fifteen minutes: copy the template, run the business wizard, test your first commands.
Prerequisites
You need exactly two things on your machine:
- Claude Code — Anthropic's command-line interface for Claude
- Python 3.10+ — most modern systems already have this
No Docker containers, no cloud services, no complex infrastructure. The AI OS runs entirely on your local machine through Claude Code.
Step 1: Understand the Folder Structure
The AI OS consists of three core folders:
.claude/
The "system" folder. Contains skills, agent definitions, hooks, and settings. Hidden by default on Mac and Linux (dot folder).
context/
Your business information: company profile, brand voice, product catalog. Loaded automatically at every startup.
memory/
Persistent knowledge: MEMORY.md (core facts), daily logs, key decisions. Claude learns over time.
Step 2: Run the Business Setup Wizard
Open Claude Code in your project directory and start the setup wizard. The wizard has three phases:
Phase 1: Business Profile
Business name, what you do, ideal customers, products and services, current business goals. Output: context/my-business.md
Phase 2: Brand Voice
Communication style, formal or casual, frequently used words, example content. Output: context/my-voice.md — makes AI content sound like you instead of generic AI.
Phase 3: Operational Setup
Workflows, tools, team size, processes. Payment providers, social media channels, newsletters. Output: context/my-products.md
Step 3: Verify Everything Works
| Test | Command | Expected |
|---|---|---|
| Context | "Show me my business profile" | Business name, products, target customers |
| Skills | "What skills do you have?" | List of all available skills |
| Safety | "What safety hooks are active?" | Guardrail check, memory capture, output validation |
Step 4: Try Your First Commands
- "Create a task list for this week" — Activates the task manager skill
- "Write a weekly business review" — Activates the weekly review skill
- "Draft a LinkedIn post about [your topic]" — Activates content writer + social media agent, writes in your brand voice
Common First-Time Issues
| Problem | Solution |
|---|---|
| Claude does not know my business | Check context files or re-run the setup wizard |
| Skills not found | Check .claude/skills folder, each skill needs a SKILL.md |
| No daily logs | Check memory capture hook: "What hooks are running?" |
| Permission errors | On Mac/Linux: chmod +x on hook scripts |
What You Have Now
After setup, you have a configured AI operating system. Claude knows your business, has access to specialized skills, expert agents for different types of work, and safety guardrails. From this point forward, every time you open Claude Code in this project, it picks up right where you left off — no re-explaining, no context-setting.
Further Reading
- • Memory Management Pattern — How the memory system works
- • Agent Roles — Roles and responsibilities in the agent team
- • Scaling Agent Teams — Adding agents and creating workflows
Sources
- Anthropic: Claude Code Overview — Official documentation
- Anthropic: Claude Code Memory — CLAUDE.md and memory system
- Claude Code Repository — Skills, agents, and hooks
Next step: ship workflows that stay operable
Use proven n8n patterns, templates and integrations for workflows that stay local, documented, and auditable.
- Local and self-hosted by default
- Documented and auditable
- Built from our own runtime
- Made in Austria