So... the current workaround that we've been using is to make every step in the task "Optional" while you are in development. This allows you to skip steps during testing. Once you have dialed in your task, you can change the required steps to no longer be optional and test again.
My other piece of advise would be to NOT use the record tool to create steps - this seems to hardcode file names into the task steps that cannot be updated later through parameter edits. This causes the step to fail on the next run when it goes looking for that old file name, unlike ModelBuilder which will use the new output from the previous step. For now, try to manually setup all your task step actions.
