Recent advancements in prediction technologies have motivated the power distribution utilities to actively utilize forecasts to minimize the impact of high-impact low frequency (HILF) events. Accurate forecasting coupled with the past operational experiences potentially enables pre-event control, aiming to minimize the impact on the system in a network with finitely large numbers of automatic and manually operated control devices. Impacts to the critical loads (CLs) are minimized when the network is appropriately reconfigured before the event strikes. However, predictions are not perfectly accurate, and given limited accuracy with ever-changing weather forecasts, control actions are required to be constantly updated until the event has passed through. Once the events are predicted with high enough confidence and the necessary reconfiguration strategy has been identified, the manually operable switches must be configured significantly ahead of event landfall to ensure the operational crews' safe return from potentially hazardous zones. Given limited resource availability and requisite manual switching operations, some loads would also have to be deenergized, which could be alleviated if those loads remain connected through remotely operable switches until minutes before the event makes the landfall. The proposed twostage control framework with a necessary mathematical formulation facilitates the same. Furthermore, the proposed framework facilitates continuous corrective action based on available lead time. The developed proactive control framework is demonstrated using a modified IEEE 123-Node system model and a realworld isolated µ-grid based on the Cordova Electric Cooperative system with superior performance results.