Is your feature request related to a problem? Please describe.

In applications like claude code, the user doesn't have to wait idly by while the agent does stuff. They can send additional messages while the agent does stuff that gets incorporated in while it operates.

Describe the solution you'd like

We need some kind of way of expressing this capability to allow implenters of frontends to know they can send additional info to an actively running agent. Like "guidable". If guideable is true, the message for a task id can be routed to the active executing task by some means.

Describe alternatives you've considered

No response

Additional context

No response

Code of Conduct

• I agree to follow this project's Code of Conduct