domains / families / agent-tooling / write-judge-prompt

write-judge-prompt

authentic

Design LLM-as-Judge evaluators for subjective criteria that code-based checks cannot handle. Use when a failure mode requires interpretation (tone, faithfulness, relevance, completeness). Do NOT use when the failure mode can be checked with code (regex, schema validation, execution tests); use `write-code-eval`. To validate an existing judge, use `validate-evaluator`.

When to use

• Designing LLM-as-Judge evaluators
• Evaluating subjective criteria
• Checking tone
• Checking faithfulness
• Checking relevance
• Checking completeness
• Handling failure modes requiring interpretation

When not to use

• Failure modes checkable with code
• Regex checks
• Schema validation
• Execution tests
• Validating an existing judge

Skill page

Source