An agent identifies dependencies and defines outputs for each subtask before combining the results. Good decomposition reduces context overload and makes failures easier to isolate, but adds coordination cost.
Task decomposition breaks a larger objective into smaller units that can be solved, checked, or delegated separately.
An agent identifies dependencies and defines outputs for each subtask before combining the results. Good decomposition reduces context overload and makes failures easier to isolate, but adds coordination cost.