We focus on how cognitive agents, which have uncertain or incomplete information about their environment, can detect when preconditions of a running plan have become unjustified. We use an ATMS to detect automatically when a plan becomes unjustified. We offer to insert a substitution plan, according to the progress of the now unjustified plan. Thus the programmer will be able to specify to one's agent what are the actions to be undertaken to correct actions made by mistake.