Stop working on Codex, need agent orchestration tool
What variant of Codex are you using?
I am using Codex less and less, and instead using a third-party orchestration tool.
What feature would you like to see?
Codex-style tools already feel like the previous generation. A single agent working in a terminal is useful, but it’s not where the real leverage is anymore. The next important layer is coding agent orchestration: multiple agents with scoped tasks, shared context, handoffs, approvals, and visibility across issues, PRs, CI, and deploys. I’d focus less on building another solo coding assistant and more on building the system that coordinates many of them.
Running multiple Codex terminals does not scale: context sharing is manual, coordination is manual, status tracking is manual, and conflict avoidance is manual. If the user has to act as scheduler, router, and project manager for every agent, the product is missing the actual orchestration layer.
There my stats
<img width="903" height="561" alt="Image" src="https://github.com/user-attachments/assets/f3945872-e436-4757-9cc1-37a2893b6d99" />
Additional information
_No response_
5 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Have you tried the Codex app? This is the client that we've designed for multi-task and multi-agent orchestration. Please give it a try and see what you think.
Hi Eric @etraut-openai
My environment is a bit complicated.
I'm using bazel, that automatically lead you cannot run several tasks on same repo (bazel caching system prevents run multiple builds to keep it consistent). I need run it in docker, that actually a way I did early days when codex has been deployed. I used it with OpenAIs web/app version where I can setup my task and pick git repo needs working on it. ~ 1y ago, it doesn't work well, can't perform well end-to-end small tasks, but now is different. Also need in MCP tools, to get task description before execution.
We're working on remote development functionality which will allow you to run the core harness in a docker container and then connect to it from the app, which runs outside of the container. Based on what you've described of your use case, it sounds like this will work well for you.
This is exactly the feature I’ve been waiting for.
I want to use the Codex app on my Mac, but run everything on my remote Linux machine. That would fit my workflow really well.
Any rough ETA on when this might be available?