Model internals

Views for the people who write the model. They show how threats are matched to components, whether listed by name or found through component properties, which helps check that each rule fires where it should. Readers looking for risk should start from the threats instead.

Component properties