Vibe Coding

Accepting AI output without deep scrutiny. Describe what you want, accept the code if it works, move on. Code quality takes a backseat—maybe because tests already cover it, or the code is throwaway, or it’ll be rewritten later.

Example: Fixing a Layout Bug

There’s a horizontal scrollbar appearing on your page. You prompt:

There's a horizontal scrollbar on the page. Fix it.

The AI tries something—maybe adding overflow-x: hidden to the body. The scrollbar disappears, but now content is clipped on mobile. You prompt again:

Now the content is cut off on mobile. Fix that too.

This loop continues. Each fix might create new issues. Eventually something works, or you give up and investigate yourself.

When It Works

Vibe coding isn’t wrong—it’s a trade-off. It works well when:

• You’re prototyping and don’t care about code quality
• The task is low-stakes or throwaway
• You have good test coverage that catches breakage
• You’re exploring unfamiliar territory and learning as you go

The risk is accumulating fixes-on-fixes that you don’t fully understand.

Often contrasted with context engineering, but they’re not mutually exclusive. Some people switch between modes depending on the task.