This article summarises part of the work done during the first two years of the SOCS project, with respect to the task of modelling interaction amongst CL-based agents. It describes the SOCS social model: an agent interaction specification and verification framework equipped with a declarative and operational semantics, expressed in terms of abduction. The operational counterpart of the proposed framework has been implemented and integrated in SOCS-SI, a tool that can be used for on-the-fly verification of agent compliance with respect to specified protocols.