@utkusen@ChatGPT i think it must be a bug, using codex app a lot since Feb, mostly on extra high reasoning, i didn’t have this issue once. Don’t know about the website experience though
@fatih@PlanetScale I've been working on an e2e sweeper, glad to see this post. Is your setup focused on fixing the flaky tests as well or just debugging them? Are these analysis results sent to somewhere or they live in the Cursor agent session?