In open multi-agent systems, agents can enter and leave the system freely. Seeking to avoid a possible chaotic scenario, we propose to monitor these systems following an approach for the contextual regulation of laws. In our solution, agent actions can be enforced by laws from four regulatory contexts: Environment, Organization, Role and Interaction. The composition of laws from these different levels permits contextual regulations in a more flexible and complete regulation for both the modeling and implementation phases. Our approach defines a top-down contextual modeling of laws, a normative ontology to formalize the regulatory contexts, and an ontology-driven rule support for law retrieval. This approach provides a straightforward method to implement regulations according to the defined regulatory contexts, yielding law evolution, contextual composition of laws, law consistency check and agent law reasoning.