Debugging becomes much easier when you give AI enough information to understand the problem.
Instead of simply pasting broken code and asking, "What's wrong?", you can tell ChatGPT what you expected, what actually happened, and what you've already tried.
These 15 prompts are designed for specific debugging situations. Replace the bracketed sections with your own code and details.
1. Find the Bug in My Code
Find the bug in this code.
Expected behavior: [WHAT SHOULD HAPPEN]
Actual behavior: [WHAT ACTUALLY HAPPENS]
Code:
[PASTE CODE]
Identify the specific problem, explain why it happens, and provide the corrected code.
Do not rewrite unrelated parts of the code.
2. Explain an Error Message
Explain this error message in simple terms.
Error:
[PASTE ERROR]Code that caused it:
[PASTE CODE]Tell me:
- What the error means
- What caused it
- Exactly where the problem is
- How to fix it
Assume I am a beginner with this technology.
3. Debug Without Rewriting Everything
Debug the following code while making the smallest possible changes.
Problem: [DESCRIBE PROBLEM]
Code:
[PASTE CODE]Preserve the existing structure and logic wherever possible.
First identify the bug. Then show only the necessary changes and explain each one.
4. Find Logic Errors
Review this code specifically for logic errors.
Expected result: [EXPECTED RESULT]
Code:
[PASTE CODE]Don't focus on formatting or style unless it affects the behavior.
Trace the important execution steps and identify where the actual behavior diverges from the expected behavior.
5. Debug Step by Step
Help me debug this problem step by step.
Problem: [PROBLEM]
Code:
[PASTE CODE]Start by identifying the most likely cause. Give me one debugging step at a time and explain what result I should look for.
Don't suggest multiple unrelated fixes at once.
6. Find Why a Function Isn't Working
This function isn't behaving as expected.
Expected: [EXPECTED BEHAVIOR]
Actual: [ACTUAL BEHAVIOR]
Function:
[PASTE FUNCTION]Analyze the function, identify the cause, and provide a corrected version.
Explain the change in simple terms.
7. Debug an API Request
Help me debug this API request.
API: [API NAME]
Expected response: [EXPECTED]
Actual response: [ACTUAL]
Request/code:
[PASTE CODE]Check the request structure, parameters, headers, authentication handling, response processing, and error handling.
Identify the most likely cause before suggesting changes.
8. Find Problems in JavaScript
Debug this JavaScript code.
Goal: [WHAT THE CODE SHOULD DO]
Code:
[PASTE CODE]Check for syntax errors, scope problems, incorrect conditions, asynchronous issues, undefined values, and incorrect DOM or API usage.
Only report issues that are relevant to the code.
9. Find Problems in Python
Debug this Python code.
Goal: [WHAT IT SHOULD DO]
Code:
[PASTE CODE]Check for syntax errors, incorrect logic, variable problems, data-type issues, exceptions, and inefficient operations.
Explain the actual cause of each important problem and provide corrected code.
10. Review Code for Hidden Bugs
Review this code for bugs that might not appear during normal testing.
Code:
[PASTE CODE]Look for:
- Edge cases
- Unexpected input
- Null or missing values
- Incorrect assumptions
- Boundary conditions
- State-related problems
For each potential bug, explain when it could occur and how to prevent it.
11. Fix a Failing Test
Help me fix this failing test.
Test:
[PASTE TEST]Code being tested:
[PASTE CODE]Failure/error:
[PASTE ERROR]Determine whether the problem is in the test or the implementation.
Explain your reasoning and provide the smallest appropriate fix.
12. Find Performance Problems
Analyze this code for performance problems.
Code:
[PASTE CODE]Problem: [WHAT FEELS SLOW]
Identify the operations most likely responsible for the slowdown.
Explain why they are inefficient and suggest practical improvements.
Don't optimize code that has no meaningful performance impact.
13. Debug Code After a Recent Change
This code worked before I made the following change.
Previous code:
[PASTE OLD CODE]New code:
[PASTE NEW CODE]Current problem:
[DESCRIBE PROBLEM]Compare the two versions and identify the change most likely responsible for the problem.
Explain why it caused the issue and show the smallest fix.
14. Create a Minimal Reproduction
Help me reduce this bug to a minimal reproducible example.
Original code:
[PASTE CODE]Problem:
[DESCRIBE PROBLEM]Remove unrelated code while preserving the behavior that causes the bug.
Return:
- The minimal example
- An explanation of what triggers the bug
- The information I should provide when asking others for help
15. Perform a Final Debugging Review
Perform a final debugging review of this code.
Goal: [WHAT THE PROGRAM SHOULD DO]
Code:
[PASTE CODE]Check for:
- Runtime errors
- Logic errors
- Edge cases
- Incorrect assumptions
- Potential data-handling problems
- Obvious performance issues
Separate confirmed problems from possible problems.
For every confirmed problem, provide the cause and the smallest practical fix.
What to Include When Asking AI to Debug
The quality of a debugging response depends heavily on the information you provide.
Whenever possible, include the code, expected behavior, actual behavior, and exact error message.
If the problem appeared after a recent change, include the old and new versions. If the issue only happens with certain input, provide an example of that input.
For example, this:
My function doesn't work. Fix it.
gives very little information.
This is much more useful:
The function should return the total price including tax. It currently returns NaN when the quantity is entered as an empty string. Here is the function and the error output: [CODE]
Clear debugging information gives AI something concrete to investigate instead of forcing it to guess.
Don't Accept the First Fix Automatically
An AI-generated fix can look convincing while still introducing another problem.
After receiving a solution, ask AI to explain why the bug occurred, test the edge cases, or review the corrected code again.
For larger projects, you can also ask it to make the smallest possible change rather than rewriting an entire function.
That makes it easier to understand what actually fixed the problem.
For more copy-ready prompts, explore the Prompt Library. For broader prompting techniques, visit the Guides section.
