What job should the planner own in a planner-executor agent?

Example Answer

The way I'd think about it is this: The planner should own deciding what work needs to happen and in what order, not silently doing the work itself. I want it to turn a messy goal into a bounded sequence of steps with dependencies, success criteria, and clear places where the system should re-evaluate.

The executor should own carrying out those steps against tools and environment constraints. If the planner starts acting like an executor, the architecture gets muddy fast because you lose separation between strategy and action.

A good planner also knows when not to over-plan. If the task is small, the best plan may be one step. The planner’s job is to create enough structure to improve execution, not to manufacture complexity.

Common Poor Answer

A weak answer is saying the planner should think and the executor should act. That is directionally true, but it does not explain planning boundaries, re-evaluation points, or how the split helps debugging.