# Deep Learn General

> Deep learning guide for any general topic: science, history, systems, concepts, soft skills, domain knowledge. Learn deeply using the Feynman technique, analogies, and applied projects. Triggers on: understand X, explain X to me, learn about X, study X deeply, how does X really work (non-coding topics).

- Skill: `yuangjay/deep-learn-general` (Agent Skill)
- Install (CLI): `npx skillmds@latest add yuangjay/deep-learn-general`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yuangjay/deep-learn-general/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: yuangjay (https://skillmd.com/u/yuangjay)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/yuangjay/deep-learn-general

---


# Deep Learn — General Topic

## Domain: General / Cross-Domain

---

## Phase 1: Frame the Topic (L1)

Before learning anything, frame what you're about to study:

1. **Write down what you THINK you already know** — 3–5 bullet points.
   (These are your pre-conceptions — some will be wrong, and that's the point.)

2. **Identify the core question the topic answers**:
   - e.g., "How does X work?" / "Why does X exist?" / "What are the principles of X?"

3. **Find the boundaries**:
   - What is IN scope (what this topic covers)?
   - What is OUT of scope (what it hands off to)?

4. **Find one authoritative source** (textbook, official guide, Wikipedia "overview" section).

**Check**: What did you think you knew that turned out to be slightly wrong?

---

## Phase 2: Build the Mental Model (L2)

1. **Read one solid, complete introduction** — not a summary listicle.

2. **Map the key concepts**:
   ```
   [Core Concept A]
         │
         ├──causes──► [Effect B]
         │
         └──requires──► [Prerequisite C]
   ```

3. **Find the best analogy**:
   - What everyday thing is this LIKE?
   - Where does the analogy break down? (That breakdown = important nuance)

4. **Identify the 3 most common misconceptions** about this topic.
   - Often found by searching "[topic] common mistakes" or "[topic] misconceptions"

**Check**: Can you draw a concept map from memory?

---

## Phase 3: Apply It (L3)

Every topic, no matter how abstract, needs application:

| Topic type | Apply it by... |
|---|---|
| System/process (e.g., how internet works) | Trace a real request end-to-end |
| Concept (e.g., game theory) | Analyse a real-world situation using the framework |
| Skill (e.g., writing, communication) | Produce something using the technique |
| Science (e.g., thermodynamics) | Solve a real problem or explain a real phenomenon |
| History/event | Write a structured analysis: causes → events → effects |

**Output**: Produce something concrete that demonstrates the knowledge.
This could be:
- A written explanation (blog post, summary doc)
- A diagram or visual
- A working prototype or experiment
- A structured analysis or argument

---

## Phase 4: Stress-Test the Understanding (L4)

1. **Find counter-examples**: Where does this concept break down or not apply?
2. **Find expert disagreement**: What do experts argue about within this topic?
3. **Connect to other domains**: How does this topic relate to something you already know well?
4. **Teach it**: Explain to the AI. Use no jargon — if you use a term, define it first.

---

## Active Recall Prompts

After each session, answer from memory:

1. "Summarise [topic] in 3 sentences."
2. "What is the most important insight about [topic]?"
3. "What is something most people get wrong about [topic]?"
4. "How would you explain [topic] to a curious 12-year-old?"
5. "What's one real-world example of [topic] in action?"

---

## Feynman Test

> "Explain [topic] to me from scratch. Assume I know nothing.
> Use an analogy. Tell me where the analogy breaks down."

---

## Spaced Repetition

- **+1 day**: Write a 5-sentence summary from memory
- **+3 days**: Explain to the AI without notes — identify remaining gaps
- **+1 week**: Find one expert source (book, lecture, paper) and deepen one part
- **+1 month**: Teach it to someone else or apply it to a new problem