codex still keeps trying to do minimal implementations of stuff even if explicitly told not to in AGENTS.md and prompts
Resolved 💬 2 comments Opened Sep 23, 2025 by FreddyE1982 Closed Nov 11, 2025
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
I have been adding very extensive rules in AGENTS.md AND prompts to ensure codex does full implementations of everything. But it will still try implementing "minimal versions" of classes, functions or entire features...saying stuff like...
"I´ll implement a minimal version of...", "I´ll code a stub of...", "I`ll create a minimal implementation...", "Will do minimal x of y for now..."
As long as I have to always check the protocols if codex did something "minimal" its not really as useful and helpful as it could be.
2 Comments
Split the task into parts. This is because you are asking the model way too complex implementations.
Thanks for the bug report. We've shared the feedback with the team members who are responsible for training our codex models. We're tracking this feedback and similar reports so we can make the model more robust in future iterations.
If you see other model behavior that you'd like to report, please use the recently-added
/feedbackcommand in the CLI. That will provide us with additional details including logs and conversation text.