task-management

You can run asynchronous background tasks for an autonomous AI agent by starting bounded iterations via start_background_task, then using check_task_status and get_task_result to monitor execution and retrieve outputs.

You can schedule periodic system checks or scheduled reports by applying the cron_expr scheduling mode, or use every_seconds and at modes to orchestrate time-based background execution and track status.

Managing long-running analyses without blocking responsiveness is best handled by delegating them to sub-agents as bounded background tasks, allowing the main agent to continue operating while tracking task status.

Yes, you can cancel scheduled background tasks using the cancel_task operation and list all active background tasks via list_background_tasks to maintain clear visibility and lifecycle control.

The limitation of using bounded background tasks is that iterations are constrained to maintain agent responsiveness, meaning tasks must operate within defined lifecycle boundaries rather than executing indefinitely without status checks.

Yes, task-management supports one-off scheduled tasks for periodic checks through the at scheduling mode, providing clear status handling and lifecycle management from start to result retrieval.