Codex Desktop: coordinate voice and text so the same task is not run twice
Open 💬 2 comments Opened Jul 29, 2026 by iqbalbhawana
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
Summary
Codex Desktop voice and the text chat interface can appear to operate as separate task streams. If a user asks for the same implementation through both, it is unclear whether the work is shared, duplicated, or independently executed.
Why this matters
Users naturally expect voice to be another interface for the same Codex task. Without a clear shared-state indicator, the same request can be sent twice, potentially causing duplicate work, conflicting changes, unnecessary usage, or confusing status updates.
Requested behavior
Please provide clear task ownership and coordination between voice and text in Codex Desktop. For example:
- Show whether voice and text are attached to the same active task.
- Detect substantially duplicate requests across active voice and text sessions and warn before starting a second execution.
- Route a duplicate request to the existing task when appropriate.
- Make it obvious when a separate task will be created.
- Consider allowing only one execution channel to control a task at a time, while the other remains available for clarifications and sharing logs or screenshots.
Observed context
This was observed on macOS in Codex Desktop while using voice alongside the Desktop chat interface. No application version was recorded.
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Thank you for flagging #35500. I reviewed it and am keeping this issue
open because the reports are related but distinct.
#35500 concerns preserving context when an existing text task
transitions into live Voice. This issue concerns coordination when
voice and text can operate as separate active task streams: users need
clear task ownership, duplicate-request detection, and protection
against the same implementation running twice or producing conflicting
work.
A unified text-to-Voice transition may reduce the chance of
duplication, but it does not replace explicit coordination when
separate voice and text tasks remain possible.