Based on code review: - import modules, not classes - use methods, not lambdas for the mock command prompt class - use None rather than Mock for DUMB_SORT - autouse two fixtures and remove them from test signatures
Validate update_completion sets the completions widget's model correctly based on the command text.