Add ability for MCPs to provide images to Codex
Resolved 💬 2 comments Opened Sep 16, 2025 by aehlke Closed Nov 11, 2025
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
What feature would you like to see?
As we can drag images into Codex, I also need my MCP to be able to provide an image to it.
Example use case: MCP that runs my app in Xcode and takes a simulator screenshot to debug and fix.
Are you interested in implementing this feature?
_No response_
Additional information
_No response_
2 Comments
Thanks for filling this. I haven't checked if there is any issue for it. For reference https://github.com/openai/codex/discussions/2085#discussioncomment-14414424
It looks like it works, but Codex does not gain any knowledge from image content.
Instead, it fills up the context really fast, which makes me think that it forwards the base64 content.
<img width="499" height="415" alt="Image" src="https://github.com/user-attachments/assets/76962b51-ca5f-46ff-950f-f6bd531b2cbb" />
Thanks for the feature request. This was recently added. See this issue for details.