DEV Community

Cover image for I Turned 12 Classic Thinking Frameworks into AI Skills — One Command, Structured Reasoning
peterfei
peterfei

Posted on

I Turned 12 Classic Thinking Frameworks into AI Skills — One Command, Structured Reasoning

#productivity #ai #beginners #opensource

TL;DR: Built an open-source tool that distills 12 classic methodologies (First Principles, Systems Thinking, Five Forces, etc.) into structured AI prompts. One command to install, then just say "analyze this with First Principles" and your AI follows the actual methodology step-by-step. GitHub

Here's the thing I noticed using AI tools daily:

Ask ChatGPT to analyze a business idea, and it gives you a decent answer. But it's generic. It doesn't know how you want it to think.

You could spend 10 minutes writing a prompt explaining Porter's Five Forces framework. Or... what if you could just say "analyze this industry with Five Forces" and the AI already knows exactly how?

That's what I built.

What it does

For each methodology, I went back to primary sources (not blog posts explaining blog posts). Original books. Original papers. Original talks by the creators. Then extracted 3-7 core principles (each backed by >=2 cross-domain proofs) and built an executable protocol the AI follows.

The result: 12 "skill files" that you install with one command:

npx skills add peterfei/forge-skill-first-principles-skill
Enter fullscreen mode Exit fullscreen mode

Then just talk naturally:

"Walk me through a First Principles analysis of why SpaceX succeeded with reusable rockets"

And the AI doesn't give you a generic response. It actually:

  1. Separates known facts from assumptions
  2. Reduces the problem to fundamentals
  3. Reasons upward from those fundamentals
  4. Checks for analogy traps

The 12 methodologies

Thinking tools:

  • First Principles (Aristotle → Descartes → Musk)
  • Systems Thinking (Meadows, Senge, Forrester)
  • Occam's Razor (William of Ockham → modern ML)
  • Feynman Learning Technique (teach to learn)

Business & Strategy:

  • Lean Startup (Ries, Blank — BML loop)
  • Growth Hacking (Ellis, Chen — AARRR, north star metric)
  • Porter's Five Forces (industry profit analysis)
  • Kaizen (Toyota Production System — 1% daily improvement)

Design & Innovation:

  • Design Thinking (IDEO, Stanford d.school)
  • Double Diamond (British Design Council)

Problem Solving:

  • TRIZ (Altshuller — 40 inventive principles from 40K patents)
  • Premortem (Gary Klein — assume failure, work backwards)

Why this matters

LLMs are general reasoning engines. Their default mode is "be helpful with whatever context you have." But experts don't just know more facts — they apply structured thinking frameworks that have been battle-tested across decades.

Every one of these skills passes a 6-point quality check:

  1. ✅ 3-7 core principles with cross-domain evidence
  2. ✅ Executable step-by-step protocol
  3. ✅ Clear "when NOT to use this" boundaries
  4. ✅ >=3 misuse patterns with detection logic
  5. ✅ >=3 honest limitations (no BS disclaimers)
  6. ✅ >50% primary sources

Try it

Everything's MIT licensed. Works with Claude Code, Cursor, Codex CLI, Cline, and 13+ other AI runtimes.

# Pick what you need
npx skills add peterfei/forge-skill-first-principles-skill
npx skills add peterfei/forge-skill-systems-thinking-skill
npx skills add peterfei/forge-skill-five-forces-skill
npx skills add peterfei/forge-skill-premortem-skill
# ... or all 12
Enter fullscreen mode Exit fullscreen mode

GitHub Repo

The engine itself is also open-source — you can forge your own methodologies if you want something not in the 12.

What methodology am I missing? I'm collecting ideas for the next batch. Drop suggestions in the comments or open an issue on GitHub.

Top comments (0)