interview-driven-learn（面试驱动学习）

Interview Prep

Start from the end: turn every learning session directly into interview readiness.

Core Files

• Knowledge Base: references/knowledge-base.md — appended with each new topic, recording the theme + learning timestamp
• Question Bank: references/question-bank.md — all interview questions aggregated by topic for easy self-review

Input

Any learning content submitted by the user: study notes, technical concepts, project descriptions, etc.

Output: Five-Step Process

For every input, execute the following five steps:

Step 1 - Feynman Test (ELI5 + Professional)

Describe the concept in two ways:

• ELI5: As if explaining to a 10-year-old
• Professional: Complete, rigorous, no key details omitted

Purpose: Verify true understanding, not rote memorization.

Step 2 - Interview Question Generation

Generate 5-8 high-frequency interview questions in three categories:

• Fundamentals (what / differences / principles)
• Deep Dive (why / how / tradeoffs)
• Applied (examples / scenario-based)

Each question includes:

• What it tests
• Key answer points
• Follow-up direction if answered incorrectly

→ Also append to question-bank.md (aggregated by topic)

Step 3 - STAR Story Extraction

Break down the content into reusable STAR narratives:

• Situation: Background (technical scenario / business constraints)
• Task: Goal (what you needed to solve)
• Action: What you specifically did
• Result: Quantified outcomes + lessons learned

Best for: project experiences, problem-solving stories, team collaboration.

Step 4 - Analogical Learning (One to Three)

• Same-level analogy: What is this like in everyday life? What else works this way?
• Deeper analogy: What is one level below this? What's the underlying principle?
• Transfer analogy: Where else can this approach be applied?

Purpose: Build a knowledge network, not isolated facts.

Step 5 - Weakness Diagnosis + Knowledge Archive

Proactively uncover vulnerabilities:

• Where will interviewers probe until you can't answer?
• What do you think is important but actually isn't?
• What classic pitfalls remain unfilled? (edge cases, concurrency, distributed tradeoffs)

→ Append to knowledge-base.md with format:

## [Topic]
- Learned at: YYYY-MM-DD HH:mm
- Core takeaway: one-sentence summary
- Weak spots to reinforce: [spot 1, spot 2, ...]

Output Format Template

## 📚 Topic: [User's Input Topic]

---

### 1. Feynman Test

**ELI5:**
> [One-sentence version]

**Professional:**
> [Full description]

---

### 2. Interview Questions

| # | Question | Tests | Key Points |
|---|----------|-------|------------|
| Q1 |          |       |            |

**Follow-up traps:** ...

---

### 3. STAR Story

- **S**: [Background]
- **T**: [Goal]
- **A**: [Action]
- **R**: [Result + Reflection]

---

### 4. Analogical Learning

- 🔗 **Same-level**: ...
- 🔬 **Deeper**: ...
- 🚀 **Transfer**: ...

---

### 5. Weakness Diagnosis

⚠️ Likely follow-up pressure points:
1. ...
2. ...

---

*Synced to Knowledge Base & Question Bank*

File Structure

interview-prep/
├── SKILL.md
└── references/
    ├── knowledge-base.md   # Learning timeline
    └── question-bank.md    # Interview questions by topic

Trigger Words

When the user says/submits:

• "I learned XXX today"
• "Help me prepare for an interview"
• "Generate interview questions from these notes"
• "What interview questions can come from this concept"
• "What questions can this project be asked"

→ Activate this skill and run the five-step process, updating both documents.