AI writes confident code. Confidence is not correctness.
Direct answer
Review AI-generated code the way you would review a new colleague's pull request: read every change, run it, test it, check edge cases and security, and make sure it fits your project's style. Ask the AI to explain anything you do not understand, and do not merge code you cannot explain.
A checklist
- Does it solve the actual problem? Compare with your request.
- Run it. Do not trust code that has not run.
- Read the diff line by line.
- Look for hidden changes: extra files, removed code, changed config.
- Check edge cases: empty input, null, very large input.
- Check errors: are failures handled sensibly?
- Check security: secrets, input validation, authentication, queries.
- Check dependencies: are new packages needed and real?
- Write or run tests that would fail if the code were wrong.
- Check style and naming against your project.
Common AI mistakes
- Calling functions or packages that do not exist.
- Out-of-date API usage.
- Over-engineered solutions.
- Passing tests by weakening them.
Where Teyro fits
Teyro's Working with AI Coding Agents course turns this into short lessons you can play in a few minutes a day. It has 60 lessons, and you can start free.
The bottom line
You are responsible for the code you ship, whoever wrote it.
