An adaptive agent changes its behavior in response to the changes it observes in its operating environment. Conversely, it exhibits a particular behavior in between adaptations. It is beneficial for other agents to know when an agent is engaged in one kind of behavior rather than another to plan their activities. To these ends, an agent can offer a specification of its behavior, as well as when and how that behavior may change. Altogether, forming a promise, i.e., an indication of what the behavior might be when a follow-up request reaches the agent. However, promises can have different temporal horizons. The shorter the horizon, the more communication events are needed to keep other agents informed about the (perspective) behavior of an agent, and more re-planning needs to take place. In this work, we explore what such promises might look like by drawing inspiration from long-running research in temporal logics. Doing so permits us to leverage existing techniques for program synthesis from temporal specifications for planning.