Class Judge

Private Readonly _aiConfig

Private Optional Readonly _logger

Private Readonly _runner

Private Readonly _sampleRate

Private _buildEvaluationInput

• _buildEvaluationInput(input, output): string

• Builds the evaluation input string passed to the runner.

  Combines the original prompt and the response into a single, well-known format the judge model is expected to evaluate.

  Parameters

  • input: string

  • output: string

  Returns string

Private _getEvaluationMetricKey

• _getEvaluationMetricKey(): undefined | string

• Gets the evaluation metric key from the judge AI config. Treats empty strings and whitespace-only strings as invalid.

  Returns undefined | string

  The evaluation metric key, or undefined if not available

Private _parseEvaluationResponse

• _parseEvaluationResponse(data): undefined | {
      reasoning: string;
      score: number;
  }

• Parses the structured evaluation response. Expects top-level {score, reasoning}. Returns score and reasoning, or undefined if parsing fails.

  Parameters

  • data: undefined | Record<string, unknown>

  Returns undefined | {
      reasoning: string;
      score: number;
  }