How should source lookup count when evaluating an agent?

Distinguish reasoning from inputs, source-assisted answers, and what each result supports.

This is a question I would like to explore. Votes help me understand readers’ interests; the writing may evolve with the research.