Clair's list of dependent applications seems to be increasing and it's hard for Clair maintainers to test the performance implications and the veracity of the results by hand (as is currently the case).

To negate future headaches it is important to invest time and effort into developing frameworks for testing:

• Load testing: How the code performs under load?
• Acceptance testing: How accurate are the results based expected outcomes?

These tests should be continually running based on a (TBD) schedule, and should alert the maintainers on failure.