You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Codes CLI: Actions Executed Prematurely and Ignored Instructions #6878
The CLI often performs actions before the current task is completed, even when I specifically instruct it not to. For example, I’ve asked several times that it avoid running certain steps early - like tsc and git while we’re still working on UX polishing — but it still does so.
In addition, there are cases where the CLI does not respond to direct prompts or instructions, which makes the workflow feel inconsistent and difficult to control. By contrast, Claude behaves much more friendly and predictably and is easier to work with.
It would be great if the CLI could respect task sequencing more reliably and respond consistently to user instructions.
Have the control of when to run certain commands like git and tsc.
Be more talkfull. Say what it thinks before coding.
Additional information
Currently, I can work either with a plan or with an API key.
I think it would be useful to have a hybrid mode:
after the plan is exhausted, Codex could notify the user and suggest to switch to the API key for the next X hours.
Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can make future models more robust and efficient. And thanks for using the /feedback command. That additional information is useful.
Activity
etraut-openai commentedon Nov 20, 2025
Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can make future models more robust and efficient. And thanks for using the
/feedbackcommand. That additional information is useful.