AI Corrected Feynman

Run the AI-corrected Feynman technique on a concept — the user explains it in their own words, Claude finds real errors/vague spots/unsupported causal claims (not encouragement), scored 1-10, repeated until it scores 10 — then logs every round into the concept's own note. Supports a chain mode that walks an ordered, categorized list of related concepts one at a time, and a review mode that schedules spaced repetition (SM-2) on already-passed concepts. Use when the user wants to "用AI纠错费曼学习法学一下 X", "费曼一下 X", practice explaining a concept for AI correction, work through a "学习链"/concept chain step by step, asks what's due for review / wants to review passed concepts, or asks to log/continue such a session.

goodluckz Updated 1 repo stars

File contents

goodluckz/claude-code-config commit 2c285e90e8

Frequently asked questions

npx get-skill add goodluckz/ai-corrected-feynman