Codex Monitor now fully supports the Codex collab feature; it can spawn any number of new agents at…
@Dimillian63K followers
26,955 views
33 Comments
Comments from the original post on X (Twitter)

@victor_wu
Currently, the only bug that makes me feel this software is not quite perfect is that after I select the model's tier, for example, if I select xhigh, and then switch to another window and come back, it will revert to medium.❤️
3 likes

@Dimillian
Yeah I’ll fix that soon

@victor_wu
But I can't wait to promote you to the Chinese community! I've already consumed 6.2 billion tokens on Codex CLI, and I'm very clear about the pros and cons of using Codex CLI. Thank you for bringing such an unparalleled piece of software. By the way, were you a PM or a programmer before?
1 like

@Dimillian
Mostly a software engineer but also doing a lot of other things. I’m always building my own apps
1 like

@victor_wu
Frankly, you understand user needs better than those programmers who use Codex CLI. Thank you again for developing such wonderful software.
1 like

@Dimillian
Thanks!!

@DaleLJefferson
Really nice. Maybe they should show up nested under the initial agent or rendered inline like how Claude Code does it.
3 likes

@Dimillian
They could but only for the current session, for now the graph don’t really return the information
1 like

@DaleLJefferson
They are full new agents you seem to be able to ask follow up prompts this is going to unlock new workflows.

@Dimillian
Exactly, where is a whole list feature where one threads can have a list of attached agents but it seems not fully implemented yet

@rudrank
Wow
2 likes

@moehajje
soo good!
1 like

@DaleLJefferson
What I don’t understand is the Codex team have 10+ developers latest models and unlimited tokens yet their first party tools are missing features and clunky. It’s not a great showcase for Codex that they can’t use it to build a decent product.
1 like

@Dimillian
I think they're scaling operations based on how active the Codex repo is. Also, the truth is that sometimes one individual coordinating multiple agents goes faster than multiple humans + agents. I think big corpo will need a lot of engineering flow refinements.
4 likes

@victor_wu
Hi, I tested it myself and found that I can't enable the subagent. I also used the exact phrase you used in the video, so is there something I should set up beforehand?
1 like

@Dimillian
Did you enabled it in the experimental settings?

@victor_wu
I found my mistake. I enabled this feature after the conversation started, so I either need to start a new conversation or exit and re-enter the current conversation to make it work. I'm so grateful for this feature! My biggest complaint about CodeX CLI before was this, and you've completed the final piece of the puzzle.

@madanlalit68
Orchestration as a prompt is the dream. Just hope the agents don't start a union in the UI.
1 like

@VanCoyld
What is Codex Collab feature ? Basically like subagents? I thought subagents were not working with Codex
1 like

@AgenticPM
Very nice design system
1 like










