Add ability to set up a parallel session to a currently running codex session

Resolved 💬 3 comments Opened Mar 26, 2026 by saulrichardson Closed May 10, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What variant of Codex are you using?

codex cli/codex app

What feature would you like to see?

I would like to be able to have on codex session doing the work and then i have another one that is basically the same session but where I can ask it questions about why it did what it did or what other ways things could have been done. Basically, I'd like to be able to easily have a second agent that can easily monitor another agent (see everything it does and all the reasoning behind it) but where I can interrogate it and then leave other agent to implementing. The workflow right now is implementation, then interrogation, and then refactor but if i could somehow interrogate the agent while it is working then it would speed things up and make real time observability much easier. I think this could work by having the mirror session be a modified fork of the initial session but where it can still access the current state. It could just spectate or even interrupt the session if the user decides to do so.

tldr; adding ability a codex session to have parallel sessions that can monitor the active thread could make observability and develop easier and quicker.

Additional information

_No response_

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 5 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #14188
  • #15482
  • #14164
  • #14722
  • #15137

Powered by Codex Action

saulrichardson · 5 months ago

Doesn't appear to be a duplicate of the flagged issues as this feature request is more concerned with setting up a parallel agent that still has access to the active thread and can give developer ability to interrogate the current session with an agent rather than waiting for the implementation to finish and also make observability more powerful.

etraut-openai contributor · 3 months ago

This feature request hasn't received enough upvotes, so I'm going to close it.