every bug below is one your ai coding assistant will ship to you. most of them, more than once. each entry names the bug, shows where it shows up, and links to a runnable chapter on the fix when one exists.
send a bug to @TFisPython. the best ones land here, anonymized.
why:asked to filter `None` from `[0, 1, None, 2]`, the ai writes `[x for x in xs if x]` and silently drops `0`. shipped a `user_id` outage at one company.
fix:use `if x is not None` for nullable checks. never use truthiness for nullable id checks.
why:writes `def add(item, items=[]):` and the same list gets reused across every call. classic python footgun, ai writes it weekly.
fix:use `items: list | None = None` and create the list inside the function.
why:writes `if items == None:` six times in one file. `__eq__` can lie. `is` cannot.
fix:use `is None` always. it's not a style choice.
why:ask for the top 3 best sellers, get the 3 worst. `sorted()` is ascending by default and the ai never asks.
fix:always pass `reverse=True` when you want largest-first, or use `heapq.nlargest`.
why:asks for a streaming generator, gets a function that builds the full list in memory and returns it. memory blows up at scale.
fix:look for `yield`. no yield, not a generator.
why:`for k in d:` then `del d[k]` raises `RuntimeError: dictionary changed size during iteration`. ai writes it confidently anyway.
fix:iterate over `list(d.keys())` if you need to mutate while looping. or build a new dict.
why:writes `try: ... except Exception: pass` and hides real bugs forever. the call site reports success.
fix:catch the specific exception. if you need a wide net, log + reraise. never silent.