I would go from the most pressing problem, apply the Pareto principle to solve it, and repeat until reasonably happy. What is the biggest actual pain that the c

I would go from the most pressing problem, apply the Pareto principle to solve it, and repeat until reasonably happy. What is the biggest actual pain that the current lack of docs causes? I do not yet know the answer (feel free to share), but there is a good chance that the problem does not actually require a formal description of every screen one by one. You may be trying to create reference docs where higher level and more abstract documentation is warranted.