Cleaning of young forest stands is a multicriteria problem with conflicting goals. This kind of forest management treatment is performed by human beings but it is possible that this work may be performed by artificial agents in the future. The artificial agents need detailed information about how to clean a forest stand and/or what are the goals for cleaning. One problem in development of cleaning rules for artificial cleaning agents is that explicit knowledge about good cleaning results is not detailed. In this paper we present a tool for developing and testing rules and judging evaluation functions for cleaning. We illustrate this tool by presenting examples of some ways to clean forest stands in a computer environment and we present how the cleaning results can be evaluated. In order to obtain material for experiments it is also possible to simulate forest stands using this tool.