바이브할 때와 이해할 때
작업마다 적절한 신뢰 수준을 선택합니다.
바이브할 때와 이해할 때은(는) CoddyKit의 무료 Vibe Coding 강의입니다. 이것은 4개 중 1번째 강의입니다. 아래에서 전체 강의를 무료로 읽을 수 있으며, 내장 코드 에디터와 24/7 AI 튜터와 함께 브라우저에서 직접 실습할 수 있습니다. 이 강의는 Vibe Coding 학습 경로의 일부이며, 진행 상황이 웹과 CoddyKit 앱에 동기화됩니다. Vibe Coding 강의에는 총 4개의 강의가 포함되어 있습니다.
이 강의의 일부는 아직 번역되지 않았으며 영어로 표시됩니다.
Two Modes of Building
Vibe coding feels like magic: you describe what you want, the AI writes it, and it just works. But great builders know there are really two modes of working with AI.
Vibe mode: you trust the AI to produce something, you run it, and you iterate by feel. Understand mode: you slow down, read the code, and make sure you know exactly what it does. The skill isn't picking one forever, it's knowing which mode each task deserves.
This lesson teaches you that judgment.
Why Blind Trust Bites You
If you vibe through everything, you eventually hit a wall: a bug you can't fix, a security hole you didn't see, or a feature the AI keeps breaking because you can't tell it precisely what's wrong.
AI tools like Cursor, Claude Code, and Copilot are incredible accelerators, but they don't understand your intent the way you do. They pattern-match. When the stakes are low, that's fine. When the stakes are high, you need to understand what shipped.
The Stakes Question
Before each task, ask one question: what happens if this code is silently wrong?
- A typo in a landing-page headline? Low stakes, just vibe it.
- Logic that charges a customer's card? High stakes, understand it.
- Code that deletes user data? High stakes, understand and test it.
The answer to the stakes question sets your mode.
Green Zone: Pure Vibe
Some work is perfect for full vibe mode. These are reversible, low-risk, and easy to eyeball:
- UI layout, colors, spacing, copy
- Throwaway prototypes and demos
- One-off scripts you run once
- Boilerplate and config you can regenerate anytime
In v0, Bolt, or Lovable, you can generate a whole landing page from a sentence. If you don't like it, regenerate. Cost of being wrong: near zero.
A Green-Zone Prompt
Here's the kind of prompt you can fire off without reading every line of output. You're describing an outcome and trusting the tool to fill in details.
Build a hero section for a coffee subscription startup:
- Big headline, one-line subheadline, an email signup box
- Warm cream background, dark brown accents
- Mobile-friendly
If I don't like it, I'll just ask you to regenerate.Red Zone: Understand First
Other work demands understand mode. These are hard to reverse and dangerous when wrong:
- Authentication and login flows
- Payments and billing logic
- Anything that deletes or overwrites data
- Permission checks (who can see what)
- Database migrations
Here you still use AI, but you read every line, ask it to explain itself, and test the edge cases before shipping.
Make the AI Explain Itself
In understand mode, don't just accept code, interrogate it. After the AI writes something sensitive, ask it to walk you through the logic in plain English. This is the single best habit for staying in control.
You just wrote the function that checks if a user can delete a post.
Explain it line by line in plain English.
Then list every case where the WRONG user could slip through.
Don't change the code yet, just explain.A Tiny Bit of Real Code
Understand mode doesn't mean you become a CS professor. It means you can read the shape of the logic. Look at this permission check, then run it. Notice how easy it is to read even if you couldn't write it from scratch.
function canDelete(user, post) {
return user.id === post.authorId || user.role === 'admin';
}
const alice = { id: 1, role: 'user' };
const post = { authorId: 2 };
console.log(canDelete(alice, post)); // false: not her post
console.log(canDelete({ id: 9, role: 'admin' }, post)); // true: adminThe Middle Zone
Most real work lives in a yellow middle zone: a form, a search filter, a data fetch. Here you vibe the first draft, then spot-check the risky parts.
Strategy: generate fast, verify selectively. Let the AI produce the whole feature, then point your attention only at the lines that touch money, data, or access. Skim the rest.
Build Your Own Trust Ladder
Effective builders keep a mental trust ladder:
- Trust fully: visuals, copy, prototypes
- Trust then skim: forms, fetches, filters
- Trust nothing until explained: auth, payments, deletes, permissions
Over time you'll move tasks up the ladder as your own understanding grows. That's the whole point: AI lets you build now and understand more each cycle.
Decide Out Loud
A simple ritual: before you accept an AI's code, say out loud which mode you're in. "This is just copy, full vibe." Or "This touches the database, understand mode."
Naming the mode forces the stakes question and stops you from sleepwalking risky code into production. It takes three seconds and saves hours of debugging.
Quick Check
Which task most clearly belongs in "understand mode" rather than full vibe mode?
Recap
You learned to match your trust to the stakes:
- Ask "what if this is silently wrong?" before every task.
- Green zone (visuals, copy, prototypes): full vibe.
- Red zone (auth, payments, deletes, permissions): understand first, make the AI explain itself.
- Middle zone: generate fast, verify the risky lines.
Knowing which mode to use is the core skill of building wisely with AI. Next, we'll look at the specific ways AI code goes wrong, so you can catch it.
자주 묻는 질문
“바이브할 때와 이해할 때” 강의는 무료인가요?
네 — “바이브할 때와 이해할 때” 전체 내용을 이 웹사이트에서 무료로 읽을 수 있습니다. 인터랙티브하게 실습하려면(내장 코드 에디터와 24/7 AI 튜터), CoddyKit PRO로 업그레이드하면 Vibe Coding 강의 전체를 잠금 해제할 수 있습니다. Vibe Coding 강의에는 총 4개의 강의가 포함되어 있습니다.
“바이브할 때와 이해할 때”에서 뭘 배우나요?
작업마다 적절한 신뢰 수준을 선택합니다. 브라우저에서 직접 실행하는 실습 코드로 Vibe Coding을(를) 배우며, 24/7 AI 튜터가 강의를 진행하면서 질문에 답변해줍니다.
Vibe Coding을(를) 시작하는 데 경험이 필요한가요?
사전 경험은 필요하지 않습니다. CoddyKit의 Vibe Coding은(는) 초급자부터 고급 학습자까지를 위해 구성되어 있으므로, 여기서 시작하거나 처음부터 시작할 수 있으며 자신의 속도대로 진행할 수 있습니다. 이것은 4개 중 1번째 강의입니다.
“바이브할 때와 이해할 때” 강의는 얼마나 걸리나요?
대부분의 CoddyKit 강의는 약 5~10분이 소요됩니다. 각 강의는 간결하고 인터랙티브하여 꾸준한 진행이 가능하며, 웹과 앱에서 중단한 부분부터 바로 시작할 수 있습니다.
이 Vibe Coding 강의에서 코드를 작성하고 실행할 수 있나요?
네. 모든 Vibe Coding 강의에는 내장 코드 에디터가 포함되어 있으므로, 브라우저에서 바로 실제 코드를 작성하고 실행한 후 즉시 AI 피드백을 받을 수 있습니다 — 로컬 설정이 필요 없습니다.
이 강의의 모든 강의
- 바이브할 때와 이해할 때
- 인공지능 코드의 함정 피하기
- 진정한 개발자로 성장하기
- 나만의 바이브 코딩 플레이북