s4h · skills-for-humanity

The methods layer for AI reasoning.

  knowledge   ·   what it knows    ·   training
  tools       ·   what it can do   ·   MCP · APIs
  methods     ·   how it thinks    ·   s4h        ◄── the missing layer

Models gave AI knowledge. MCP gave it tools. s4h adds the missing layer — methods: history's most rigorous thinking, rebuilt as procedures Claude runs.

---

Prompt vs. method

Ask a raw model	Run the method
"Should I leave my stable role for the startup?"	/s4h-decision-reversibility-analysis
One balanced-sounding pass, shaped entirely by how you framed it.	Classifies the decision by reversibility → finds the closing window → applies the rigor that class warrants.

Same model. A different quality of thought.

---

The library

5 clusters · 34 domains · 200+ methods

Think Sharper	/logic /probability /decision /constraint /game-theory /epistemology /investigation /economics /information /cognition
Think Differently	/creativity /analogy /play /evolution /design
Think About People	/communication /social /emotional /ethics /identity /narrative /psychology /mindset /writing /linguistics
Think in Time & Systems	/systems /temporal /historical /resource /strategy /network /ecology
See More Clearly	/aesthetic /sensory

Commands are prefixed s4h- — e.g. /s4h-decision. Each domain routes to 4–20 individual methods.

---

Why a methods layer matters

• Repeatable — the same rigor on any problem, not whatever got improvised.
• Inspectable — you see the moves, not a black box.
• Composable — methods chain: a decision feeds a premortem feeds an incentive map.
• Democratized — the toolkit of history's finest thinkers, on demand.

---

Install & use

npx @human-avatar/skills-for-humanity

/s4h

Describe your situation — it routes to the method. You never need to know which one. Every skill states its read and asks how deep to go before it runs; correct the framing anytime.

Lineage · de Bono · Meadows · Altshuller · Minto · Goldratt · Tetlock · Klein · Mill · Kant · Aristotle · Noddings · Rawls · Sun Tzu · Clausewitz · Musashi · Nash · Axelrod · Vickrey · Shapley · Spence · Marshall · Coase · Ostrom · Granovetter · Barabási · Shannon · Norman · Lakoff · Grice · Kahneman · Tansley · Dawkins

---

See it think — worked examples

Why social media makes discourse worse despite everyone wanting better

Platforms want healthier discourse. Users say they want it. It keeps getting worse. /s4h-systems-feedback-mapping finds the reinforcing loop: engagement-optimised feeds reward content that provokes reaction, which selects for users who engage with provocation, which trains the algorithm to serve more of it. The stated goal and the actual incentive structure point in opposite directions. No individual decision made it this way — the system produced it.