I also hate reviewing my own ai generated code. Worse; it’s impossible for me to keep focus. So I created a skill to keep it interesting. It’s nothing more than a prompt that asks the ai to ask me questions about the code. Starting at high level, like ‘Describe the core responsibilities of X, Y, and Z in a sentence or two. From that it goes deeper. It is also instructed to summarise based on my answers. If I give a wrong, or vague answer it will flag it as code that might be problematic.
Mostly, this helps to keep me focused and really understand the code. It’s not perfect. On two occasions a secondary human review found a critical bug, like calling a non-user-facing endpoint (but with similar name).
Mostly, this helps to keep me focused and really understand the code. It’s not perfect. On two occasions a secondary human review found a critical bug, like calling a non-user-facing endpoint (but with similar name).