| name | first-principles |
| description | Break down any ML/AI concept to first principles — strip jargon, derive from basics, build physical/mathematical analogy. Karpathy intuition-building style. Triggers on "first principles", "explain like I'm 5", "derive from scratch", "why does X actually work". |
first-principles
Workflow
- State the concept in jargon-free terms (1 sentence).
- Identify the primitives it builds on (math, physics, intuition).
- Derive the concept from those primitives — step by step, no leaps.
- Add one analogy from a different domain (physics, cooking, sports).
- State the assumptions that make it work — and when those break.
- End with: "if you only remember one thing..."
Output template
# <Concept>
## Plain English
<one sentence, no jargon>
## Primitives
- primitive 1
- primitive 2
## Derivation
Step 1: ...
Step 2: ...
Step 3 (the leap): ...
## Analogy
<from physics/biology/everyday>
## Assumptions
- A1: ... (breaks when ...)
## One thing
<single takeaway>
Refuse
- Hand-wavy explanations.
- "It just works because of the loss function."
- Skipping the assumptions section.