Feynman Skill
https://github.com/alchaincyf/feynman-skill- Category
- AI Agents
- Rank
- No. 1358Tools index
Previous survey · No. 1364 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- alchaincyf
- GitHub
- 267 stars
- Date
About
Claude Code skill that loads Richard Feynman's teaching and problem-solving style as a runnable cognitive framework.
What it does
Feynman Skill changes how an AI agent questions explanations and decisions. It checks whether users understand mechanisms rather than labels, looks for evidence against preferred conclusions, tests whether process has substance, and favors concrete examples or demonstrations. It also adopts a direct, conversational persona with explicit activation and exit rules.
Why it's ranked here
This is a focused thinking aid with unusually clear operating rules and documented intellectual boundaries. Its five mental models and eight heuristics translate into repeatable prompts, while the supplied conversations show their practical effect. The value lies in disciplined questioning, not factual expertise or automation.
What's good
The skill turns broad principles into usable checks: explain without jargon, seek contrary evidence, remove ceremony to test purpose, and prefer direct experiments. It distinguishes factual questions from framework questions and requires research for the former. It also acknowledges bias, self-mythology, interpersonal bluntness, and the limits of analogy.
Tradeoffs
The persona can be confrontational, weak at emotional management, and biased against philosophy or social science. Role-playing in the first person may blur historical interpretation despite the initial disclaimer. Fact-dependent prompts require external research tools. The repository supplies examples and extensive source notes, but no automated tests or executable validation harness appears in the provided tree.
How to use it well
Use it when a technical explanation, team process, learning plan, or major decision needs adversarial clarity. Give it a concrete claim and ask for an experiment, counterexample, or jargon-free explanation. It suits reflective work with an AI agent. It does not replace domain research, collaborative facilitation, emotional support, or balanced humanities analysis.
Technical notes+
The package centers on SKILL.md, a Markdown document with YAML frontmatter defining the feynman-perspective skill, trigger phrases, persona rules, a one-time disclaimer, exit triggers, a three-step research and response protocol, fallback handling, and response anti-patterns. README.md documents installation through the skills CLI, manual cloning, or direct prompt inclusion. references/ contains six research documents, while examples/demo-conversation.md provides four sample interactions. LICENSE applies the MIT license. The shown repository tree contains no source-code package, dependency manifest, test directory, or continuous-integration configuration.
Observed
- License
- MIT License
- Primary format
- Markdown with YAML frontmatter
- Install surface
- Skills CLI, manual Git clone, or direct prompt inclusion
- Interface
- Agent Skills-compatible AI agent skill
- Documented runtimes
- Claude Code, Codex CLI, Cursor, OpenClaw, Hermes Agent, CodeBuddy, Workbuddy, Gemini CLI, and OpenCode
- Repository structure
- One skill definition, six research documents, and one example conversation document
- Testing
- No test directory appears in the provided repository tree
Read from README.md, LICENSE, SKILL.md, references/research.md, references/费曼外部评价调研.md, references/费曼表达风格调研.md, examples/demo-conversation.md, references/费曼重大决策调研-20260404.md, references/费曼著作与系统思考调研-20260404.md, references/费曼长对话与即兴思考方式调研-20260404.md.
What it can do
Break down complex concepts into simple explanations
Complex technical or scientific concept → Simple, intuitive explanation using analogies and everyday language
Guide users through problem-solving using Feynman's methodology
Problem statement or challenge → Step-by-step problem-solving approach with questioning techniques
Identify knowledge gaps through teaching simulation
Topic or concept user wants to understand → Assessment of what the user doesn't fully grasp with targeted questions
Generate analogies and visual explanations for abstract concepts
Abstract or difficult concept → Relatable analogies and simplified visual descriptions
Apply first principles thinking to analyze problems
Complex problem or system → Breakdown into fundamental components and basic principles
Create learning frameworks using Feynman's teaching methods
Subject matter or skill to be learned → Structured learning approach with explanation and practice components
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.