9 Golden Rules To Set Up Claude Properly

Do these once. Benefit from better output in every conversation.

Rule 1: Download the Desktop App

Claude.ai is fine. The desktop app is better. Go to claude.com/download and install the desktop app. It gives you access to Cowork, file reading, and a smoother experience.

Rule 2: Pick the Right Model

Pick the newest Opus model from the model menu next to the send button. Model names change every few months, so choose the latest Opus in the menu. For hard work, open "Effort" in the same menu and raise it to High or above. Higher effort uses your limits faster, so keep routine tasks on the default.

Rule 3: Use Cowork, Not Chat

Cowork reads your actual files. Chat doesn't. Point Claude to your folder. It reads everything inside before responding.

Rule 4: Build Your Folder Structure

Create 4 folders inside your Cowork project:

• ABOUT ME — Who you are, your work, your best templates
• PROJECTS — Active project files
• TEMPLATES — Reusable document templates
• CLAUDE OUTPUTS — Where Claude delivers finished work

CLAUDE OUTPUTS is the only write folder. It delivers everything there. One subfolder per project.

Rule 5: Write Your "About Me" File

A .md file about who you are. What you do. How you think. Your current priorities. Claude reads it before every response. Less prompting, better output.

Rule 6: Write Your Anti-AI Style File

A .md file listing every word and pattern you refuse to see in AI output. Claude follows it. Your output stops sounding like a robot.

Examples of anti-AI words to ban: "delve", "leverage", "synergy", "unlock", "game-changer", "cutting-edge", "revolutionize", "seamless", "robust".

Rule 7: Start Asking

End every prompt with: "Ask me clarifying questions first." Claude generates a form. You answer. The output jumps 2–3 levels in quality.

Rule 8: Set Up Global Instructions

Settings > Profile > Global Instructions. Write how you write, what you hate, what you expect. Claude reads this before every single conversation.

Rule 9: Edit Your Prompt

Don't send a new message. Edit the first one. Every follow-up eats tokens. Editing keeps your context clean.

Setup Checklist

• [ ] Desktop app installed
• [ ] Newest Opus selected, with effort raised for hard work
• [ ] Cowork folder pointed at your project
• [ ] 4-folder structure created (ABOUT ME, PROJECTS, TEMPLATES, CLAUDE OUTPUTS)
• [ ] about-me.md written
• [ ] anti-ai-style.md written
• [ ] Global Instructions configured
• [ ] First prompt ends with "Ask me clarifying questions first"
• [ ] Editing prompts instead of sending follow-ups

Based on Ruben Hassid's "9 Golden Rules To Set Up Claude Properly" infographic from how-to-ai.guide