0Pricing
Vibe Coding · 강의

프롬프트 반복 개선

다시 시작하지 말고 다듬으십시오. 단계별로 인공지능을 이끄십시오.

프롬프트 반복 개선은(는) CoddyKit의 무료 Vibe Coding 강의입니다. 이것은 4개 중 4번째 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.

The First Answer Is a Draft

Beginners often delete everything and start over when the AI's first try isn't perfect. Pros do the opposite: they iterate. They treat the first output as a draft and steer it with follow-up messages.

This is the rhythm of real vibe coding — refine, don't restart. This lesson teaches you how to nudge the AI toward exactly what you want, step by step.

The Iteration Loop

Every build follows a simple loop:

  • Prompt — ask for something
  • Run — preview the result in Bolt, v0, or your editor
  • React — note what's right and what's off
  • Refine — give a focused follow-up

You repeat this loop a few times per feature. Each pass gets you closer.

Refine, Don't Restart

The AI remembers the conversation. So you don't re-describe the whole app — you just say what to change. This keeps everything you already liked.

Restarting (wasteful):
  "Build the whole todo app again but with bigger buttons."

Refining (smart):
  "Make the Add button bigger and move it to the right of
   the input."

Change One Thing at a Time

When you ask for five changes at once, it's hard to tell which fix helped and which broke something. Iterate in small steps so you can see the effect of each one.

Step 1: "Change the theme color to purple (#7c3aed)."
   (check it looks right)
Step 2: "Now add a subtle hover effect to the buttons."
   (check again)
Step 3: "Now make it work on mobile widths."

Point to What's Wrong

Good iteration feedback names the specific thing that's off and what you want instead. 'It's bad' gives the AI nothing to work with.

Weak feedback:  "That's not right, try again."

Strong feedback: "The list items are cramped. Add more vertical
                  space between them and a thin divider line."

Keep What Works

When part of the output is great, say so and protect it. Telling the AI what to keep stops it from rewriting code you already liked during the next change.

"The layout and colors are perfect — keep those exactly.
 Only change this: when the list is empty, show the text
 'No tasks yet. Add one above!' centered in gray."

Iterating on Logic, Step by Step

Iteration works for behavior too. Start with a basic version, then layer on rules. Here's a greeting function after one refinement step — try running it.

// Step 1 result: a simple greeting
// Step 2 refinement: "also handle an empty name"
function greet(name) {
  if (!name) return 'Hello there!';
  return 'Hello, ' + name + '!';
}

console.log(greet('Sam'));
console.log(greet(''));

When to Restart Instead

Iteration is the default — but sometimes a fresh start is faster. Restart when:

  • The code has become a tangled mess after many edits
  • You realize the approach is fundamentally wrong
  • The AI is stuck in a loop, re-breaking the same thing

In that case, write a clean new prompt that includes everything you learned.

Roll Back Bad Steps

Sometimes a refinement makes things worse. Don't try to patch a patch — go back. Most tools let you undo: Cursor and Claude Code track edits, and Git lets you revert to a known-good version.

A clean rollback plus a sharper prompt beats five rounds of damage control.

"That last change broke the layout. Undo it and go back
 to how the page looked before. Then instead, just
 increase the font size of the headings to 24px."

Give Feedback in the AI's Language

Speed up iteration by reacting to concrete things: the live preview, an error message, or a screenshot. Many builders like v0 and Lovable let you click an element and say 'change this'.

The more precisely you point, the fewer rounds you need.

Fast iteration cues:
- "See the red error in the console? Fix that."
- "The button I just clicked — make it full-width."
- "Match the spacing to the screenshot I shared."

A Real Iteration Session

Here's how a few turns of refinement feel in practice. Each message is short because the AI remembers the rest.

You: "Build a tip calculator: bill input, tip % buttons, total."
AI:  (produces it)
You: "Add a 'split between N people' field."
You: "Round each person's share to 2 decimals."
You: "Keep all that — just make the total bold and bigger."

Quick Check

The AI's first version of your app is 90% right — the layout is great but one button is the wrong color. What's the best next move?

Recap: Steer, Don't Restart

Great vibe coding is a conversation, not a single command:

  • Treat the first answer as a draft
  • Run the prompt → run → react → refine loop
  • Change one thing at a time, point to what's wrong
  • Keep what works; roll back bad steps
  • Restart only when the approach is truly wrong

You've now mastered the core of prompting: structure, specificity, context and iteration. You're ready to build your first real app by talking to AI!

자주 묻는 질문

“프롬프트 반복 개선” 강의는 무료인가요?

네 — “프롬프트 반복 개선” 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.

“프롬프트 반복 개선”에서 뭘 배우나요?

다시 시작하지 말고 다듬으십시오. 단계별로 인공지능을 이끄십시오. 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 4번째 강의입니다.

“프롬프트 반복 개선” 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

  1. 좋은 빌드 프롬프트의 구성
  2. 구체적으로 작성하기
  3. 인공지능에 맥락 제공하기
  4. 프롬프트 반복 개선
← Vibe Coding(으)로 돌아가기