I'll admit that I don't see what the advantage of soloist is. What I would probably do is build the runlist via include_recipe statements in pivotal_workstation::default. Then I would create a Berkfile (http://berkshelf.com/) for it, use 'berks install --path' to bundle pivotal_workstation and all its dependencies, and tar that up and upload it somewhere.

The bootstrap then becomes something like

  gem install chef
  echo {"run_list": "recipe[pivotal_workstation]"} > node.json
  chef-solo -j node.json -r http://example.com/pivotal_workstation.tar.gz

The JSON could of course be customized if you needed to set any attributes.

Co-maintainer of pivotal_workstation and soloist here. The as-yet unreleased Soloist 1.0.0, head of master on https://github.com/mkocher/soloist, uses Librarian to yank in dependencies if you have a Cheffile.

Soloist itself is just a tool to simplify generating a runlist and specifying node attributes. In fact, it just reads out of a single YAML file and does what that file says. It even gives you what might be considered role support.

Plus we totally take pull requests!