LLM-as-judge

Using a language model to score, grade, or gate another model's output against criteria, forming the automated evaluation backbone of the AI diff-review verification step.

grounded in: doctrine verification layer 1 lists 'AI diff review' as a required gate; grounds the existing ai-review + constitutional-ai + evals nodes in the durable LLM-as-judge evaluation paradigm

Connected concepts

AI diff review, Evals & benchmarks, Constitutional AI, Verification loops, Claude Fable 5