This paper presents a distributed CBR-based reasoning framework for problem-solving in a multi-agent environment. We were motivated to use preparation, adaptation and reuse in order to exploit regularities in the task domain to enhance future performance. While it is premature to state the specific benefits and limitations in our approach, we have found increased efficiency and solution quality in trial runs. At present, no invariants are guaranteed but we have attempted to maintain the efficacy of the prepared solution traces. Our techniques may not transfer directly to other domains but we expect the principles presented here will.