Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Has anyone had their own experience of how Claude or similar AI agents perform in large (1M+ lines) legacy code bases? To give a bit more context, I work on a Java code base that is 20+ years old. It was continuously updated and expanded but contains mostly spaghetti code. Would Claude add any value here?


Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: