The matrix group recognition project has reached the stage at which one can use the composition tree as a basis for further calculations. In particular, it is desirable to adapt the “trivial Fitting” paradigm, which is the basis for many successful permutation group algorithms. For this, it is necessary to obtain radical generators and a representation for the radical factor, to split the radical in elementary abelian factors, and to adapt the existing permutation group algorithms in aspects such as element tests in subgroups. Using the example of conjugacy classes of elements I will show how this can be done.