{{ message }}
Repository navigation
Prompt queuing enhancements #440
Description
What feature would you like to see?
Are there ways for some prompts/commands to be able to queue up? For instance:
- /auto will currently interrupt any existing ongoing process in the session. It'll pick up where it's left off, but I was hoping it can let the previous prompt finish running first
- /review can't be queued in Codex CLI. I use a custom text expansion that sends it in as a prompt that looks like this: "/review please review". This way of queuing works in Codex CLI, but doesn't work in Code as it still detects "/review" being used.
Activity
5ocworkshop commentedon Dec 8, 2025
I support your idea.
I think what you're saying is you're like both the current interrupt method and a way to have the prompt get picked up in the normal cycle at some future point. I'd like that to and I think that does happen, the problem is sometimes it is a very long time before the prompt gets picked up and I can't quite tell what is happening, especially if I have a lot of subagents dispatched.
Perhaps having something akin to the "thinking" traces that you can optionally expand on web chat models would be good. I know we can turn on reasoning, but that doesn't seem to be all the reasoning that is going on. The system often indicates it is off thinking or crafting a prompt etc. but there is no reasoning output to be found to understand what it is doing for multiple minutes. Perhaps it is just the slow respond of the OpenAI API in that regard, I'm not sure. But understanding when a queued prompt will get picked up would be good.
ongspxm commentedon Mar 4, 2026
https://developers.openai.com/codex/cli/features/
hmmm seems to be in the codex cli, but i tried doesnt work on jsut-ever code