Code Review ChatGPT Prompt Maker: Get Structured Feedback, Not a Wall of Text

"Review this code" is a prompt technically works, but it usually gets back a loosely organized wall of text — some real issues, some nitpicks, no indication of what actually matters most. A structured prompt fixes that before the model ever sees your code, by telling it exactly what to look for and how to report back.

What Goes Into a Structured Review Prompt

• The language — so the model reviews against that language's own conventions rather than generic advice.
• Context on what the code does — a one-line description of the function or module's purpose, which helps the model catch logic errors it otherwise couldn't tell were errors.
• Specific focus areas — bugs, security, error handling, code quality, performance, testing gaps — chosen deliberately instead of "check everything."
• A reporting format — asking for a severity level and a suggested fix per issue turns the output into something you can act on in order, rather than a paragraph to parse yourself.

An Example Generated Prompt

Picking Python, adding brief context, and enabling severity tags and suggested fixes produces something like this:

Act as an experienced Python code reviewer. Here is context on what the code does: A function that processes user-uploaded CSV files and inserts rows into a database.

Review the code below for:
1. Bugs and logic errors — incorrect logic, off-by-one errors, improper null/undefined handling, and race conditions.
2. Security vulnerabilities — injection risks, authentication/authorization gaps, hardcoded secrets or credentials, and insecure data handling.
3. Error handling and edge cases — missing error handling, unhandled exceptions, and edge cases like empty inputs, nulls, or boundary values.

For each issue found, report its location, a description of the problem, and a severity level (Critical / High / Medium / Low) and a specific suggested fix.

Code to review:
```python
[PASTE YOUR CODE HERE]
```

Everything above the code fence was generated by picking options — no prompt-writing skill required, and the same structure works whether you're reviewing five lines or five hundred.