Boris Cherny, the person who built Claude Code, says there's no secret to prompting. Andrej Karpathy says his prompts are getting shorter and dumber. And Anthropic removed over 80% of Claude Code's own system prompt for its advanced models, with no measurable loss. So I went through what they've published and pulled out the new rules. I break down how to prompt now, which effort level fits which job, how to tell Claude what done looks like, how to clean up your instructions, and three mods that get you more out of the same subscription. Download: The Claude Cheat Sheet, a copy-paste example for every rule. Attach it to a chat and Claude rewrites the examples for your job. The mods zip is here too. Attached below and in Video Downloads.

Welcome to WeAreNoCode, the free community for non-technical people who want to make money with AI. Drop a comment with three things: 1. Where you are in the world. 2. The industry you know best. 3. What you'd build if you knew someone would pay for it. Then open the Classroom and start The 8 Steps. Step 1 takes seven minutes and comes with the Four Wrong Turns diagnostic. Every video's downloads live there too. Rules are short: no promotions, no links to your own stuff, title your posts specifically, search before you ask, stay respectful. When you want the day-by-day version with questions answered live every Wednesday and the first-customer guarantee, that's WeAreNoCode Pro: https://www.skool.com/wearenocodepro/about