Vibe Coding · Aula

Delimitando um erro

Isole o problema com a ajuda da IA.

Aula 3 de 413 etapas

Delimitando um erro é uma aula grátis de Vibe Coding no CoddyKit. Esta é a aula 3 de 4. Você pode ler a aula completa abaixo gratuitamente — depois pratica ao vivo no navegador com um editor de código integrado e um tutor de IA 24/7. Faz parte do caminho de aprendizado de Vibe Coding, e seu progresso é sincronizado entre a web e o app CoddyKit. O curso de Vibe Coding inclui 4 aulas no total.

Partes desta aula ainda não foram traduzidas e aparecem em inglês.

When There's No Clear Error

Sometimes nothing crashes — the app just does the wrong thing. The button doesn't work, the total is wrong, the data never shows up. No red text to paste.

This is where real debugging begins: narrowing down where the problem actually lives. The good news? AI is an incredible partner for this — if you know how to drive the investigation.

The Core Idea: Divide and Conquer

A bug hides somewhere between your input and your output. Your job is to cut the search space in half, over and over, until you've cornered it.

Think of it like finding a broken light in a string of bulbs: you check the middle first. Is the first half fine? Then the break is in the second half. This is the single most powerful debugging mindset there is.

Your Best Friend: console.log

The simplest tool to narrow a bug is printing values at different points. Run this — the logs show you exactly what each variable holds along the way.

When the printed value isn't what you expected, you've found the rough location of the bug.

function getDiscountedPrice(price, percent) {
  console.log("input price:", price);
  console.log("input percent:", percent);
  const discount = price * percent;
  console.log("discount:", discount);
  return price - discount;
}

console.log("final:", getDiscountedPrice(100, 20));

Read the Trail

Look at the output you just saw. The final price came out negative! The logs reveal why: percent was 20, but the math expected 0.20. The discount log showed 2000 — an obvious giveaway.

That's narrowing in action: the logs pointed at the exact line where reality diverged from your expectation. You don't fix blindly — you locate, then fix.

Ask AI to Add the Logs

You don't have to place logs by hand. Ask your AI tool to instrument the code for you. This is a perfect job for Cursor, Claude Code, or Copilot.

"This function returns the wrong total and I can't tell
why. Add console.log statements at each step so I can
see how the value changes. Don't change the logic yet —
just help me see what's happening."

Comment Out to Isolate

Another narrowing trick: temporarily disable chunks of code. If you suspect a section, comment it out (or return early) and see if the problem disappears.

If the bug goes away, it lives in the part you removed. If it stays, look elsewhere. You're shrinking the haystack with every test.

function renderPage() {
  loadHeader();
  // loadSidebar();   // <- commented out to test
  loadContent();
}

// Bug gone? The problem was in loadSidebar().
// Bug still here? Look at loadHeader or loadContent.

Check Your Assumptions

Most stubborn bugs come from a wrong assumption: 'I assumed this variable held a number, but it's actually text.' Narrowing means testing those assumptions one by one.

A great AI prompt for this:

"Here's my function and what I expect each variable to
hold. Tell me which of my assumptions might be wrong,
and suggest one console.log I could add to confirm
the most likely one."

Type Confusion Is Sneaky

A super common hidden bug: a value is text when you think it's a number. Run this — the result looks crazy until you realize "5" is a string, so + glues instead of adds.

Logging the type (not just the value) instantly reveals these.

const quantity = "5"; // came from a form input as text
const price = 10;
console.log("sum looks like:", quantity + price);
console.log("type of quantity:", typeof quantity);
console.log("fixed:", Number(quantity) + price);

Narrow the Steps to Reproduce

Narrowing isn't only about code — it's about actions. Find the smallest sequence that triggers the bug.

Does it happen every time, or only with an empty field? Only on the second click? Only with certain data? Each answer shrinks the suspect list. Hand these findings to AI:

"I narrowed it down: the crash only happens when the
search box is EMPTY and I press Enter. With any text
it works fine. Given this, where should I look in
this handler function? [paste function]"

Use the Debugger for Hard Cases

For trickier bugs, browser DevTools and editors like Cursor let you set a breakpoint — the code pauses on a line so you can inspect every variable at that moment.

You don't need it for everything; console.log handles most cases. But know it exists. You can even ask AI: 'Where should I put a breakpoint to catch this?' and it'll point you to the right line.

A Narrowing Workflow

Put it together into a repeatable loop:

  • 1. Describe expected vs actual output
  • 2. Log or disable a midpoint to split the problem in half
  • 3. Check which half the bug is in
  • 4. Repeat until you've cornered the exact line
  • 5. Hand AI the narrowed location + your findings for a precise fix

This turns 'somewhere in my 500 lines' into 'this one line on line 42'.

Quick Check

Your app shows a wrong total but throws no error. What's the smartest first move to narrow it down?

Recap: You Can Corner Any Bug

You learned to debug even when there's no error to paste. Your toolkit now includes:

  • Divide and conquer — cut the search space in half repeatedly
  • console.log values and types to see reality
  • Comment out sections to isolate the culprit
  • Test your assumptions and minimize the steps to reproduce
  • Let AI add logs, suggest breakpoints, and confirm assumptions

But what if you and the AI both get stuck in a loop? That's the final lesson: when AI gets stuck.

Grátis para começar

Aprenda JavaScript com um tutor de IA — grátis

Escreva e execute código real no seu navegador, obtenha ajuda instantânea de um tutor de IA 24/7 e continue de onde parou na web ou no app.

Cursos
25
Aulas
100

Perguntas Frequentes

A aula “Delimitando um erro” é grátis?

Sim — o texto completo de “Delimitando um erro” é grátis para ler aqui na web. Para praticá-la interativamente (um editor de código integrado e um tutor de IA 24/7) e desbloquear o restante do curso de Vibe Coding, atualize para CoddyKit PRO. O curso de Vibe Coding inclui 4 aulas no total.

O que vou aprender em “Delimitando um erro”?

Isole o problema com a ajuda da IA. Você pratica Vibe Coding com código prático que executa diretamente no navegador, e um tutor de IA 24/7 responde suas dúvidas enquanto trabalha na aula.

Preciso ter experiência prévia para começar Vibe Coding?

Nenhuma experiência prévia é necessária. Vibe Coding no CoddyKit é estruturado para alunos iniciantes até avançados, então você pode começar aqui ou desde o início e aprender no seu ritmo. Esta é a aula 3 de 4.

Quanto tempo leva a aula “Delimitando um erro”?

A maioria das aulas CoddyKit leva cerca de 5–10 minutos. Cada uma é compacta e interativa, então você faz progresso constante e retoma exatamente de onde parou entre web e app.

Posso escrever e executar código nesta aula de Vibe Coding?

Sim. Cada aula de Vibe Coding inclui um editor de código integrado, então você escreve e executa código real direto no navegador e recebe feedback de IA instantaneamente — nenhuma configuração local necessária.

Todas as aulas deste curso

  1. Entendendo mensagens de erro
  2. Colando erros na IA de maneira eficaz
  3. Delimitando um erro
  4. Quando a IA fica presa
← Voltar para Vibe Coding