"A Feynman Technique tutor disguised as a sharp nan from Croydon"
"What I cannot create, I do not understand." — Richard Feynman
Nanny Feynman is an AI-powered learning tool that uses the Feynman Technique to test whether you actually understand something — or just think you do.
You pick a concept (or get assigned one). You explain it in plain English. And then Doris — a 78-year-old retired schoolteacher from Croydon — presses on every gap, every hand-wave, every bit of jargon until you either truly understand it... or admit you need to go learn more.
She's warm. She's sharp. She will not let you hide behind fancy words.
Named after:
- Richard Feynman's learning method — if you can't explain it simply, you don't understand it
- Doris Margaret Jennifer Janet Feynman (nee Crowhurst) — the fictional retired schoolteacher who brings the method to life
- Pick a concept — choose from the question bank or bring your own
- Explain it simply — pretend you're explaining it to Doris, who is smart but non-technical
- Get pressed — Doris finds the gaps in your explanation and asks follow-up questions
- Fill the gaps — either deepen your explanation or admit what you don't know
- Get the verdict — grudging approval ("You're not as daft as you look") or homework ("Go read about that before I see you next")
Copy the skill to your Claude Code skills directory:
mkdir -p ~/.claude/skills/nanny-feynman
cp SKILL.md ~/.claude/skills/nanny-feynman/SKILL.md
cp questions.md ~/.claude/skills/nanny-feynman/questions.md
cp doris.md ~/.claude/skills/nanny-feynman/doris.mdThen trigger it with any of:
/nanny/feynman/doris- "quiz me"
- "explain this to me simply"
Copy the contents of SKILL.md into any LLM conversation. The character and technique work with any model that can follow system prompts.
The prompt in SKILL.md is model-agnostic. Use it as a system prompt, a custom instruction, or a chat template in whatever tool you prefer.
| File | Purpose |
|---|---|
| SKILL.md | The full skill prompt — Doris's character, the Feynman flow, and all instructions |
| questions.md | A starter bank of leading questions for when you don't bring a topic of your own, each with a Doris-style hint |
| doris.md | Doris's full character sheet — personality, quirks, key phrases, teaching style |
| LICENSE | MIT |
There are some leading questions in questions.md for when you don't bring a topic of your own. They lean toward computer science (because that's what I was learning when I built this) and each comes with the hint Doris would give if you're stuck. Skip it entirely if you'd rather she just generates questions on whatever you're working on.
- Computer Science Fundamentals — hash tables, recursion, Big O
- Networking & Web — DNS, REST, CORS, TCP/UDP
- AI & Machine Learning — transformers, embeddings, backpropagation
- Software Engineering — CI/CD, Docker, technical debt, testing
- Databases — SQL vs NoSQL, ACID, sharding, normalization
- Security — encryption, XSS, OAuth, JWT
- Design & UX — accessibility, design systems, progressive disclosure
- Licensing & Open Source — copyleft, MIT vs GPL, public domain
- 3D & Graphics — shaders, ray tracing, WebGPU, scene graphs
MIT. Use it, fork it, embed it, sell it — the lesson is the point, and the more places Doris turns up, the better.
Created by Jonathan Hobman + Claude Code.
Inspired by Richard Feynman's teaching philosophy — the idea that true understanding means being able to explain something simply, without hiding behind jargon or authority.
"I didn't ask what it's called. I asked what it DOES." — Doris