Penetration of smart phones and localization technologies are enabling a variety of smart location-based social mobility services that can notably improve the quality life of citizens on the go. Most existing solutions are shaped from a specific set of data sources processed through predefined computational flows to provide a specific class of services, and rarely strive for generality and urban-scale goals. The contribution of this paper is to presents an architectural model for a general-purpose distributed coordination infrastructure to support the dynamic composition of a variety of smart social mobility services, by focusing on the coordination of the agents involved to address both individual mobility needs and urban-scale mobility issues. Following, it analyzes the key requirements that the implementation of such infrastructure should satisfy. Finally, to exemplify, we show how the framework can be applied to a urban ride-sharing service.