Writing code got cheap. Knowing whether it is right did not.
One written method for finding faults in code you wrote, code you inherited, and code an AI wrote for you. Every chapter ends with a lab in the Haunted Repo, a small task tracker with sixteen planted bugs, in JavaScript and Python.
One loop, practised on real bugs, until you can follow it without the book.
Reproduce the failure on demand, shrink it with git bisect, read the stack trace back to where the bad value began, and test one hypothesis at a time.
Make the smallest correct change, prove it with a test that fails before and passes after, and use logs, the debugger and curl when the bug lives outside your code.
Spot how generated code fails, review a generated change with a checklist, and use an assistant as a debugging partner without handing over the thinking.