evaluation

Records and replays SerpApi search evidence to detect changes affecting search-grounded AI agents.

Breaks model answers into claims and retrieves SerpApi evidence to evaluate how well the claims are supported.