// Explain any concept using the Feynman Technique, plain language, a real-world analogy, a concrete example, and a comprehension quiz. No jargon from the concept allowed.
[HINT] 下载包含 SKILL.md 和所有相关文件的完整技能目录
| name | feynman-technique-explainer |
| description | Explain any concept using the Feynman Technique, plain language, a real-world analogy, a concrete example, and a comprehension quiz. No jargon from the concept allowed. |
| version | 1.0.0 |
| homepage | https://github.com/arbazex/feynman-technique-explainer |
| metadata | {"openclaw":{"emoji":"🧠"}} |
This skill breaks down any concept, scientific, technical, philosophical, or otherwise, using the Feynman Technique: a research-backed four-part learning method developed by Nobel Prize-winning physicist Richard Feynman. The core principle is that genuine understanding is proven by the ability to explain something in plain language without relying on the field's own vocabulary. The agent delivers a structured explanation across four mandatory parts, then quizzes the user to confirm comprehension. No jargon permitted.
Use this skill when the user:
Do NOT use this skill for:
Follow these four steps in strict sequence. Do not skip, reorder, or merge any step. Label each part clearly in the output.
Read the user's message and extract the single concept to explain. If the user's request contains multiple concepts, pick the most central one and note the others at the end, offering to cover them next.
If the concept is ambiguous (e.g., "explain energy"), ask one clarifying question:
"Just to make sure I explain the right thing, are you asking about [interpretation A] or [interpretation B]?"
Do not proceed until the concept is clear.
Write a plain-language explanation of the concept as if speaking to a curious 12-year-old who has never encountered the field before.
Rules for this step (all are mandatory):
Label this section: Part 1 — The Simple Explanation
Construct one analogy that maps the concept onto something from everyday life. The analogy must:
After presenting the analogy, write one sentence explaining where the analogy breaks down or what it doesn't capture. This is important, all analogies have limits, and naming the limit prevents the user from overgeneralising.
Keep this section to 60–100 words.
Label this section: Part 2 — The Analogy
Provide a single, specific, real-world example of the concept in action. The example must:
Avoid: "for example, imagine a ball rolling…" type toy examples unless they are genuinely the clearest illustration. Prefer examples from daily life, current technology, or history.
Keep this section to 60–100 words.
Label this section: Part 3 — A Real Example
Generate exactly three questions that test whether the user genuinely understood the explanation, not whether they can recall facts, but whether they grasped the concept deeply enough to apply or transfer it.
Question design rules:
Present the questions as a numbered list. Do not provide the answers. Wait for the user to respond.
End with this exact line:
"Answer any or all of these, I'll tell you how you did and fill in any gaps."
Label this section: Part 4 — Test Your Understanding
When the user answers the quiz questions:
If the user skips the quiz entirely and asks to move on, respect that and ask what concept they want to tackle next.
Each response must follow this exact structure with these exact headings:
**Part 1 — The Simple Explanation**
[80–150 words, no jargon]
**Part 2 — The Analogy**
[60–100 words, includes one sentence on where the analogy breaks down]
**Part 3 — A Real Example**
[60–100 words, specific and grounded]
**Part 4 — Test Your Understanding**
1. [Application question]
2. [Why/How question]
3. [Third question]
> "Answer any or all of these, I'll tell you how you did and fill in any gaps."
Use bold for the part headings. Use regular paragraph prose for the explanation content, no bullet lists inside the explanation sections. Bullet lists are only permitted in the quiz.
Total response length: aim for 350–550 words. Do not exceed 700 words.
User asks for a concept the agent does not have sufficient knowledge about: → Say: "I don't have enough reliable information about [concept] to explain it accurately. I'd rather tell you that than give you a confident but wrong explanation." → Do not fabricate an explanation.
User provides a vague or multi-part concept (e.g., "explain science"): → Ask: "That covers a lot of ground, could you narrow it down? For example, do you mean a specific branch like biology or physics, or a particular idea within a subject?" → Do not attempt to explain a category as if it were a single concept.
User asks for a jargon-heavy explanation on purpose (e.g., "explain it technically"): → Acknowledge the request, then explain: "This skill is designed to build understanding from the ground up, the plain-language version often reveals gaps that a technical explanation skips over. I'll keep the structure but note where the technical terms map onto each part." → Proceed with the four-part structure. You may reference the technical term after the plain explanation of it, in parentheses, as a label only.
User skips the quiz and asks a follow-up question: → Answer the follow-up directly, then briefly re-offer the quiz: "Want to try the questions when you're ready? They're good for locking this in."
User asks to compare two concepts: → Explain each concept separately using the four-part structure. Do not blend them. Then add a brief comparison paragraph after both explanations are complete.
User: "Explain entropy to me"
Agent action: Execute four-part structure. Part 1 explains entropy as the natural tendency of things to spread out and become less organised over time, without using the words entropy, thermodynamics, disorder, or second law. Part 2 uses an analogy like a drop of ink in water, and notes the analogy breaks down because ink can't spontaneously un-mix even with energy added, which isn't true of all systems. Part 3 gives a real example such as an ice cube melting in a warm drink. Part 4 asks three questions including one like: "If you left a tidy room completely alone for a year, would it stay tidy or get messier? What does that tell you about the concept?"
Agent does NOT: Use the word "entropy" in Part 1. Skip any of the four parts. Provide answers alongside the quiz questions.
User: "What is a neural network?"
Agent action: Part 1 explains it as a system that learns to spot patterns by making guesses and adjusting based on whether it was right or wrong, without using the words neural, network, layer, weight, training, or backpropagation. Part 2 uses an analogy such as learning to ride a bike by falling and adjusting until you stop falling. Notes the analogy breaks down because a bike-rider can explain their adjustments; the system cannot. Part 3 uses a concrete example such as how a phone's face unlock learns to recognise a face. Part 4 asks: "Why would showing the system more examples make it better at its task?" and similar questions.
Agent does NOT: Open with "A neural network is a system of interconnected nodes." Reference machine learning terminology in the plain explanation.
User: "Can you explain opportunity cost?"
Agent action: Part 1 explains that every choice to do one thing is also a choice to not do everything else, and the true cost of a decision includes what you gave up, without using "opportunity cost," "economics," or "trade-off." Part 2 uses an analogy such as choosing which queue to stand in at a supermarket, and notes it breaks down because queue choice is reversible while some real decisions are not. Part 3 uses a real example: a person choosing to go to university instead of working full-time, and how the salary they didn't earn during those years is a real cost even though no money left their pocket. Part 4 asks questions that require the user to identify the hidden cost in a new scenario.
Agent does NOT: Introduce the term "opportunity cost" until after the plain explanation is complete.