feynman-tutor

# Feynman Tutor

**An AI teaching Skill based on the Feynman Technique — instead of explaining concepts to you, it makes YOU explain them, exposing the exact gaps in your understanding.**

Works with [Claude Code](https://docs.anthropic.com/en/docs/claude-code) and any LLM agent framework that supports custom System Prompts + file I/O + shell execution.

[中文版 README](README_ZH.md)

---

## The Problem It Solves

You've probably experienced this:

- You read a deep article and thought "I get it" — but when someone asks you to explain, you can't
- You asked AI to "teach you" something, it gave a perfect explanation, you nodded along — but nothing actually stuck
- You've learned lots of fragments, but they don't connect into anything usable

The problem isn't that AI explains poorly. The problem is: **passively listening ≠ learning.**

Feynman Tutor's core design: **it doesn't give you answers — it makes you articulate them yourself.** Where you can't articulate, that's where you don't understand.

---

## How Is This Different From "Ask AI to Explain"?

|  | Typical AI Chat | Feynman Tutor |
|---|---|---|
| **Who's talking?** | AI explains to you | **You explain to AI** |
| **Finding blind spots** | You don't know what you don't know | AI probes with follow-up questions to expose cracks |
| **Difficulty** | Either too easy or too hard | Always one step beyond your knowledge boundary (ZPD) |
| **Cross-session memory** | Starts from zero every time | Remembers what you've learned, where you got stuck, which analogies worked |
| **External materials** | "Summarize this video for me" | Turns the video into teaching material, uses Feynman method to help you internalize |
| **Cross-topic connections** | Each topic is an island | Automatically discovers structural analogies and shared patterns across topics |

---

## Core Features

### 1. Role-Reversal Teaching

You're not the student — you're the "teacher." Feynman Tutor plays a curious friend who knows nothing, and asks y