STAND WITH UKRAINE
codingOpenai

Code Review Prompt (OpenAI / ChatGPT)

This variant is optimised for the OpenAI API, using the system/user separation and markdown tables that GPT-4o renders particularly well in the ChatGPT interface. The table format makes issues easy to scan and copy into a GitHub PR comment. The summary section provides a quick decision signal for reviewers.

Prompt Template
## System
You are an expert senior software engineer. Respond in structured markdown.

## User
Review this **{{language}}** code:

**Context:** {{context}}

```{{language}}
{{code}}
```

## Required Output Format

### 🐛 Bugs & Logic Errors
| Line | Issue | Fix |
|------|-------|-----|

### 🔒 Security Issues
| Line | Issue | Fix |
|------|-------|-----|

### ⚡ Performance
| Line | Issue | Fix |
|------|-------|-----|

### 📖 Readability & Style
| Line | Issue | Fix |
|------|-------|-----|

### ✅ Summary
- Overall quality score: /10
- Critical issues: N
- Recommended action: [merge | revise | rewrite]

Variables

{{language}}Programming language (e.g., TypeScript, Python, Go)
{{code}}The code to review
{{context}}What the code is supposed to do

Example

Input
language: Python
context: Flask endpoint that processes webhook payloads
code:
@app.route('/webhook', methods=['POST'])
def webhook():
    data = request.json
    if data['type'] == 'payment':
        process_payment(data['amount'], data['user_id'])
    return 'OK'
Output
### 🔒 Security Issues
| Line | Issue | Fix |
|------|-------|-----|
| 3 | No signature verification on webhook | Validate HMAC signature from headers before processing |
| 4 | KeyError if 'type' missing | Use `data.get('type')` with fallback |

### ✅ Summary
- Overall quality score: 3/10
- Critical issues: 2
- Recommended action: revise

Related Tools

AI Prompt BuilderAI API Request BuilderAI Token Counter

FAQ

How is this different from the generic code review prompt?
This version uses markdown tables optimised for ChatGPT's renderer and adds an overall quality score. The generic version uses a prose format that works better in plain-text environments.
Can I use this with the OpenAI API directly?
Yes. Split the prompt at the ## User marker and send the System section as the system message and the User section as the user message in your API call.
Does GPT-4o mini produce good code reviews?
GPT-4o mini is sufficient for style and simple logic issues but misses subtle security vulnerabilities. Use GPT-4o or GPT-4 Turbo for security-sensitive code.

Related Prompts

Code Review Prompt

This prompt structures code reviews into five clear categories so the AI produces actionab...

Code Review Prompt (Claude / Anthropic)

Claude responds especially well to XML-structured prompts because its training aligns with...

Security Code Audit Prompt

Security audits require a systematic approach that covers every vulnerability category, no...