I would continue to work on the spec (actually I started today with the validator) and we have to admit that there are some tools that allow bioboxes to be interchangeable just to a certain degree. It does not mean that we should stop following the aim of creating interchangeable tools but maybe this means that we have to find way on the long ther to make it as easy as possible to use such tools. e.g:
Find a way that the tools could report in yaml which databases they need (I think in the initial propasal you wrote that he tools could report their types for example.)
... we can instead specify a list of morphisms and each container can list which of those they implement. ...
Maybe to create a container that checks if another container needs a database and downloads it and places it somewhere (something like an adapter). I mean if you want to use binning tools that uses such databases you would have to download it anyway.
I'm sure for profiling tools we will have the same problems.