Handle User Input While an Agent Is Mid-Task
Beginner
The problem, from first principles
A real-time agent doesn't have the luxury of finishing one task before hearing from the user again -- messages can arrive while it's still working. Some of those messages genuinely need to interrupt what's happening ("stop, forget it"); most should just wait their turn.
The task
Write handle_interrupt_input(current_task, new_input, interrupt_keywords). If new_input contains any interrupt keyword (case-insensitive), return "INTERRUPT_AND_SWITCH" regardless of whether a task is running. Otherwise, return "START_NEW_TASK" if there's no current task, or "QUEUE_AFTER_CURRENT" if there is.
Keywords
production-systemsstreamingreal-time-agents
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.