The usage of already available software agents plays a vital role in the process of development of application specific software. This reuse also leads to software development cost benefits as well as may ensure the timely delivery. This paper lay down an idea that for reusing reactive multi-agents systems two factors are to be considered i.e. (i) abstract description of agent in application independent way and (ii) reuse of such systems through adoption in specific domain. For such a development main requirement is the effective reusable software abstractions. In present study the role of abstraction level and dependence level is analyzed for intelligent agents.