It really depends on the interaction between the discipline of your team and the failure modes of the tool. A culture of code review which puts a high bar of justification and documentation-for-new-contributors on monkey patching can counteract this failure mode.

Do you code review what each new library does? It doesn't have to be code you wrote to cause problems. And most people treat a library as a black box with inputs and outputs. Which works well in a sane language, but often not ruby.