Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
CandidateResponse, EvaluationRubric -> EvaluationScore
Prompt a high-capability judge model with a candidate response and a structured rubric to compute evaluation scores.
Problem it solves
Deterministic string matches cannot accurately evaluate open-ended or subjective task responses.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.