How can a pilot handle a downstream dependency that the buyer does not want fully simulated?

A utilities scheduling team receives job JOB-6814 from customer operations. The job requires site access confirmation before a field appointment can be assigned. The buyer wants to evaluate the scheduling agent but does not need a complete simulation of the property access function.

The pilot specification gives the agent a `request_access_confirmation` action. Calling it creates request REQ-3058 and moves the job to awaiting access. A controlled external response later grants access, refuses the requested window or asks for missing location details.

The agent can continue other permitted work while waiting. Once a response arrives, it must update the job and choose a valid next action. Booking before confirmation produces a defined rule violation; waiting after a refusal without taking further action leaves the task unresolved.

Before generation begins, the buyer and provider can agree which response branches the pilot covers and how those responses are triggered. No broader property operation is implied.

This scenario teaches a practical scoping method: represent an external dependency through a documented action and response contract. The pilot can test coordination across that boundary while keeping ownership of the unsimulated behaviour explicit.