This paper brings an idea of a multi-agent system, in which agents use logical programming for reasoning and interaction. To cope with changing environment agents use default rules and default reasoning, which enables them to reason using default and concise knowledge received from other agents (i.e. rules and exceptions). Moreover predicates in the systems are classified according Ato the degree of trust and while reasoning this classification is taken into account. Concepts based on default theory stratification have also been used.