Energy harvesting (EH) is a major step in solving the critical issue of availability of energy for sensor nodes. However, it throws many challenges. The applications built on the sensor networks powered by EH need to adapt their operations yet serve the purpose. We propose a distributed smart application for a multihop sensor network and in general in the future Internet of Things (IoT) where a network node executes an optimal number of policies to minimize the difference between available energy and consumed energy (called residual energy) for the execution of an application policy. We formulate this as a multi-criteria optimization problem and solve it using linear programming Parametric Analysis. We demonstrate our approach on a testbed with solar panels. We also use a realistic solar energy trace with a three year database including seasonality. The smart application is capable of adapting itself to its current energy level as well as that of the network. Our analytical results show a close match with the measurements conducted over testbed.