So at what point does a full window start to hurt? You will hear a number, so here is how much weight it can bear.
The dumb zone
Also called Smart zone, Context rot
The informal claim that a model gets worse past some point in its window, so you should stay below it. That reliability falls as context grows has support. Any specific percentage is a working rule, because no one has shown us a measured line.
The habit it teaches is good: clear early, and keep noise out. But don't repeat a number as fact, least of all in an interview. Windows went from 200K to 1M, and a percentage and a token count now give very different answers. Watch for symptoms in your own sessions: rereading files, forgetting rules.
Say it to your agent
“If you reread a file you already read, or break a rule in AGENTS.md, stop and tell me. When I see /context pass 100k, I will ask you to write the state to docs/handoff.md and start a new session. That number is our rule of thumb, not a known limit.”