I've noticed my review habits slipping. When a model spits out a 200-line diff that passes tests, I skim and hit approve. Last week that bit me: a subtle off-by-one in pagination that the test suite never covered. Curious how others handle this. Do you set a hard rule like "no merging code I can't explain
Do you actually read AI-generated code before merging?
Replies
No comments yet — be the first to share your thoughts.