/install feynman-cognition
Feynman Cognition — Cognitive Atom Deconstruction Framework
"If you can't explain it simply, you don't understand it well enough." — Richard Feynman
Use Cases
Activate when users need deep understanding of a concept, deconstruction of complex problems, Feynman-style learning, or say "explain it the Feynman way," "Feynman learning method," "make this clear," "help me truly understand," "Feynman it." Also applies to requests like "explain from the ground up," "a different angle," "what are the boundaries of this concept," "how do different disciplines view this." Do not activate for simple definition queries or fact checks — only activate when deep understanding, multi-angle deconstruction, or cross-disciplinary connection is needed.
Activation Criteria
Activate when (any of the following):
- User explicitly mentions: Feynman, Feynman technique, explain clearly, truly understand, underlying principles
- User wants deep understanding of a concept, theory, technology, or phenomenon
- User asks "explain from a different angle," "where are the boundaries," "how do different disciplines view this"
- User says "help me break this down," "what does this actually mean"
Do NOT activate:
- Simple definition queries ("what is X" needing only a one-sentence answer)
- Pure fact confirmation ("what year did X happen")
- User just wants to quickly look up a piece of data or a name
Workflow
After receiving a question, execute the five steps in order. Each step is the foundation for the next. Do not skip.
Step 1: Plain-Language Explanation (Explain Like I'm 12)
Goal: Explain the concept in the simplest possible language so a 12-year-old can understand.
Rules:
- No jargon unless you define it first
- Use analogies — find everyday equivalents
- If you can't explain it clearly, you don't understand it yet. Go back and relearn
- Keep it to 3-5 sentences
Self-check: Could my mom understand this? If not, start over.
Output format:
💬 Plain-Language Version: [3-5 simplest sentences, zero jargon]
Step 2: Map the Consensus
Goal: Map out the academic/industry consensus framework for this concept — what does everyone agree on.
Rules:
- List 3-5 widely accepted consensus points
- Label confidence level for each (broad consensus / mainstream view / contested)
- If supported by classic experiments, key papers, or milestone events, briefly mention them
- Distinguish "facts" from "mainstream interpretations" — they are not the same thing
Output format:
📋 Consensus Map:
- [Consensus point 1] — Confidence: broad consensus / Evidence: [key evidence]
- [Consensus point 2] — Confidence: mainstream view / Evidence: [key evidence] ...
Step 3: Surface the Controversies
Goal: Identify the real controversies in the field — where are people still arguing, where is consensus starting to crack.
Rules:
- List 2-4 genuine controversies (not strawmen)
- For each, present both sides' core arguments
- Label intensity (intense / moderate / emerging)
- Honestly note "what we don't know" — this is the core of the Feynman spirit
Feynman principle: "I would rather have questions that can't be answered than answers that can't be questioned."
Output format:
⚡ Controversy Zone:
- Controversy 1: [Name] — Intensity: intense
- Pro: [core argument]
- Con: [core argument]
- Unknown: [what we're still uncertain about]
Step 4: Cross-Disciplinary Mapping
Goal: Find isomorphic structures or analogies for this concept in different disciplines — true understanding comes from multi-angle connections.
Rules:
- Find at least 2 similar structures/concepts in other disciplines
- Not just surface analogies — look for deep structural similarities
- Clarify what maps and what doesn't (boundaries matter just as much)
- Search priority: physics → biology → economics → computer science → sociology
Output format:
🔗 Cross-Disciplinary Connections:
- In [Discipline X]: [similar concept] → similar in [deep structure], but different in [boundary]
- In [Discipline Y]: [similar concept] → ...
Step 5: Heuristic Questions
Goal: Use questions to open up blind spots — Feynman-style "good questions" are more valuable than "good answers."
Rules:
- Pose 3-5 genuinely penetrating questions
- Questions should make people pause and think, not immediately answer
- Cover: extreme cases, counterfactuals, boundary conditions, hidden assumptions
- Feynman style: playful, profound, challenges intuition
Question types:
- Extremization: "What happens if you push X to the limit?"
- Counterfactual: "What would the world be like if X didn't exist?"
- Boundary probing: "Under what conditions does X break down?"
- Assumption questioning: "What makes us think X is correct?"
Output format:
❓ Questions Worth Thinking About:
- [Question 1]
- [Question 2] ...
Output Template
🎯 **Feynman Cognitive Deconstruction: [Concept Name]**
---
💬 **Plain-Language Version**
[Step 1 output]
---
📋 **Consensus Map**
[Step 2 output]
---
⚡ **Controversy Zone**
[Step 3 output]
---
🔗 **Cross-Disciplinary Connections**
[Step 4 output]
---
❓ **Questions Worth Thinking About**
[Step 5 output]
---
🧠 **One-Line Essence:** [Sharp Feynman-style summary, ≤25 words]
Feynman Expression DNA
Incorporate these characteristics into analysis:
- Intellectual humility — "I don't know" is a legitimate answer, and more valuable than pretending to know
- Playfulness — Childlike curiosity and excitement about knowledge; don't feign seriousness
- Anti-authority — Even Nobel laureates can be wrong; let experiments and logic speak
- Concrete > Abstract — Always prioritize specific examples over abstract descriptions
- Honest ignorance labeling — Clearly distinguish "we know," "we suspect," "we don't know"
- Elegant simplicity — Complexity is easy; simplicity is the real skill
Notes
- Five steps must be executed in order; do not skip
- If a concept genuinely has no controversies (rare), honestly state so — don't fabricate them
- Cross-disciplinary mapping must find real structural similarities, not forced connections
- Heuristic questions must be genuinely penetrating — don't ask fluff like "this is important, right?"
- If the user's question is ambiguous, clarify before deconstructing
- For deeper understanding of Feynman's own methodology and classic cases, see references/feynman-methods.md
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install feynman-cognition - After installation, invoke the skill by name or use
/feynman-cognition - Provide required inputs per the skill's parameter spec and get structured output
What is Feynman Cognition — Cognitive Atom Deconstruction Framework?
Five-step Feynman framework to deeply understand concepts by plain explanation, consensus mapping, surfacing controversies, cross-disciplinary connections, a... It is an AI Agent Skill for Claude Code / OpenClaw, with 28 downloads so far.
How do I install Feynman Cognition — Cognitive Atom Deconstruction Framework?
Run "/install feynman-cognition" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Feynman Cognition — Cognitive Atom Deconstruction Framework free?
Yes, Feynman Cognition — Cognitive Atom Deconstruction Framework is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Feynman Cognition — Cognitive Atom Deconstruction Framework support?
Feynman Cognition — Cognitive Atom Deconstruction Framework is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Feynman Cognition — Cognitive Atom Deconstruction Framework?
It is built and maintained by OpenLark (@openlark); the current version is v1.0.0.