New Feature Request: Ability to have claude code wait for external input

I have multiple claude code agents running in parallel. Sometimes I want one of the agents to wait until another agent complete a certain task and publish a message to Kafka that another agent should read from. Once the second agent reads and completes the task it will publish a message to Kafka that the first agent reads and starts working on.

The timeouts that exist currently prevent this from working this way. Would be very useful if you can add a mechanism to make this type of workflow work.