Agent image-gen tool: no reference image input / no way to reproduce or iterate
What version of Codex CLI is running?
N/A
What subscription do you have?
N/A
Which model were you using?
gpt-4o-mini? (image gen agent)
What platform is your computer?
N/A
What terminal emulator and version are you using (if applicable)?
N/A
What issue are you seeing?
Problem
In the Codex Agent image generation tool there is no way to provide reference images or reuse a seed. That means the output is not reproducible and it is very hard to iterate incrementally on an image—every run starts from scratch.
Without references, workflows like maintaining style, composition, layout, or branding become extremely painful.
What steps can reproduce the bug?
- Open Codex Agents image generation tool.
- Try to supply a reference image (e.g., upload or paste a URL) or reuse a seed.
- There is no UI to do this, so results cannot be replicated and iteration is frustrating.
What is the expected behavior?
Provide a reference image input (upload/paste/link) and allow seed reuse so an image can be reproduced and iterated incrementally.
Additional information
This is primarily a feature request for Codex Agents image generation rather than a CLI bug, but it currently makes iterative image design workflows extremely slow.
This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗