The notion of context is well-known and relevant to several research areas such as language, philosophy, logic, and artificial intelligence. In these areas, contexts are typically used to model the effect of the environment (in its most general sense, including the spatial and temporal interpretation of the term) on the communication occurring amongst active entities, such as humans or artificial agents. In the area of computer science, contexts have often be studied as a means to encapsulate knowledge (whether factual or procedural) or beliefs, and to combine them incrementally and dynamically to model many different notions related to agent knowledge, beliefs, and communication. 

Generalising upon the notion of context-dependent coordination, in this paper we introduce the notion of emph{agent coordination context} as a means to model and shape the space of agent interaction and communication. From a theoretical perspective, agent coordination contexts can serve the purposes of (i) enabling agents to model the environment where they interact and communicate (the subjective viewpoint), and (ii) providing a framework to express how the environment affects interpretation of agent communication acts (the objective viewpoint). From an engineering perspective, the notion of agent coordination context enables in principle agents to perceive the space where they act and interact, reason about the effect of their actions and communications, and possibly affect their environment to achieve their goals. Also, agent coordination contexts allow engineers to encapsulate rules for governing applications built as agent systems, mediate the interactions amongst agents and the environment, and possibly affect them so as to change global application behaviour incrementally and dynamically. To illustrate the issues discussed in this paper, we provide examples based on different coordination models and languages for agent systems, focussing in particular on Internet agent coordination.