Evaluation of the performanceMachine learning and accuracy of machine learningMachine learning Interatomic Potentials software packages has been hampered by the difficulty in using these packages in a simple workflow. We outline the problems we have faced and give recommendations for improving their usability. We identify the factors and metrics that need to be made clear in the provision of such software packages in the hope of setting standards for meaningful evaluations of machine learningMachine learning Interatomic PotentialsInteratomic Potentials software and present an illustrative example of such a comparison for three such packages.