Persistent Callback orchestration/Loop
Open 💬 2 comments Opened Mar 9, 2026 by northwoodstrader1
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What variant of Codex are you using?
CLI
What feature would you like to see?
Why does codex not have this same feature claude code has:
- a persistent local agent process that stays alive between turns
- a scheduler/job controller that can sleep, watch processes, and resume work
So the short version is:
- Claude Code likely has a persistent orchestrator around the model
- this environment is turn-based
- background Bash can monitor things here, but it cannot inject a new assistant turn by itself
basically if I start running a script I want to be able to have a bash timer set for lets say 3 minutes to auto start a next process or check the log of the script running to know if it has to fix something, an error pops up to auto fix the bug, or to auto start the next process when the script finishes. Claude code does this by using a bash task with a timer where it can do tasks in teh background, when the background timer finishes it notofies claude to check on said task or do a specific task in that item without using tokens while it essentially sleeps
Additional information
_No response_
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
This overlaps with #15679, which tracks the narrower in-chat
/loop <interval Xm> <message>implementation.Working implementation and reviewable diff are already available there:
That implementation queues recurring prompts in the same Codex thread, replaces the previous schedule when
/loopis run again, and automatically removes the loop when a loop-triggered turn determines the work is already complete.