Mobile agents are a recent paradigm to facilitate the design and programming of distributed applications. However, whilst their popularity continues to grow, a uniform theory of mobile agent systems is not yet sufficiently elaborated, in comparison with classical models of distributed computation. In this paper we show how to use mobile agents as an alternative model for implementing distributed local computation rules. In doing so, we approach a general and unified framework for local computations which is consistent with the classical theory of distributed computations based on graph relabeling systems.