Motivation
Eigent tasks can run for a long time and often require user input while the desktop app is still the execution host. Today, users need to stay near the desktop app to monitor progress, answer agent questions, or send follow-up instructions. If they step away, the task can stall or the user loses visibility into what is happening.
Remote Control makes long-running desktop work easier to manage by letting the user create a secure temporary link for an active task, then monitor and guide it from another device.
Solution
Add a first version of Remote Control for active desktop sessions.
The user should be able to create an expiring remote link from the desktop app. The remote page should show session status, live task/step updates, and allow basic control commands such as replying to human-in-the-loop questions, sending follow-up messages, skipping/stopping work, and updating the active task/project target when needed.
The desktop app remains the execution host. The remote page only sends scoped commands through a bridge, and the bridge reports delivery/acknowledgement/failure state back to the remote session.
Alternatives
One alternative is to keep the current desktop-only workflow and ask users to return to the app whenever input is needed. This is simpler, but it makes long-running tasks much less useful.
Another alternative is full remote desktop or screen sharing. That is much broader than what we need right now and introduces more security and UX complexity.
A third alternative is to rely on external messaging tools for notifications, but that would not provide reliable task targeting, command acknowledgement, or project/session context.
Additional context
Initial implementation should focus on safe command relay, session lifecycle, and clear remote state:
- expiring links
- revoke/extend support
- bridge online/offline state
- authenticated owner access plus link token
- command status: pending, delivered, acknowledged, failed, expired
Motivation
Eigent tasks can run for a long time and often require user input while the desktop app is still the execution host. Today, users need to stay near the desktop app to monitor progress, answer agent questions, or send follow-up instructions. If they step away, the task can stall or the user loses visibility into what is happening.
Remote Control makes long-running desktop work easier to manage by letting the user create a secure temporary link for an active task, then monitor and guide it from another device.
Solution
Add a first version of Remote Control for active desktop sessions.
The user should be able to create an expiring remote link from the desktop app. The remote page should show session status, live task/step updates, and allow basic control commands such as replying to human-in-the-loop questions, sending follow-up messages, skipping/stopping work, and updating the active task/project target when needed.
The desktop app remains the execution host. The remote page only sends scoped commands through a bridge, and the bridge reports delivery/acknowledgement/failure state back to the remote session.
Alternatives
One alternative is to keep the current desktop-only workflow and ask users to return to the app whenever input is needed. This is simpler, but it makes long-running tasks much less useful.
Another alternative is full remote desktop or screen sharing. That is much broader than what we need right now and introduces more security and UX complexity.
A third alternative is to rely on external messaging tools for notifications, but that would not provide reliable task targeting, command acknowledgement, or project/session context.
Additional context
Initial implementation should focus on safe command relay, session lifecycle, and clear remote state: