# LlmAsAJudgeOptions

Options for configuring an LLM-as-a-Judge custom evaluator.

Uses a foundation model to assess agent performance based on
custom instructions and a rating scale.

- Source: [`aws-bedrockagentcore/lib/evaluation/evaluator-config.ts`](https://github.com/aws/aws-cdk/blob/v2.275.0/packages/aws-cdk-lib/aws-bedrockagentcore/lib/evaluation/evaluator-config.ts#L31)
- Type last changed: v2.255.0 · 19 May 2026

## Example

```ts
// Create a custom LLM-as-a-Judge evaluator
const evaluator = new agentcore.Evaluator(this, 'MyEvaluator', {
  evaluatorName: 'my_custom_evaluator',
  level: agentcore.EvaluationLevel.SESSION,
  evaluatorConfig: agentcore.EvaluatorConfig.llmAsAJudge({
    instructions: 'Evaluate whether the agent response is helpful and accurate.',
    modelId: 'us.anthropic.claude-sonnet-4-6',
    ratingScale: agentcore.EvaluatorRatingScale.categorical([
      { label: 'Good', definition: 'The response is helpful and accurate.' },
      { label: 'Bad', definition: 'The response is not helpful or contains errors.' },
    ]),
  }),
});

// Use the custom evaluator in an online evaluation configuration
new agentcore.OnlineEvaluationConfig(this, 'MyEvaluation', {
  onlineEvaluationConfigName: 'my_evaluation',
  evaluators: [
    agentcore.EvaluatorSelector.builtin(agentcore.BuiltinEvaluator.HELPFULNESS),
    agentcore.EvaluatorSelector.custom(evaluator),
  ],
  dataSource: agentcore.DataSourceConfig.fromCloudWatchLogs({
    logGroupNames: ['/aws/bedrock-agentcore/my-agent'],
    serviceNames: ['my-agent.default'],
  }),
});
```

## Properties

- `additionalModelRequestFields`: { [key: string]: any } — Additional model-specific request fields.

  **Default:** No additional fields

- `inferenceConfig`: [EvaluatorInferenceConfig](/cdk/api/v2/docs/aws-cdk-lib.aws_bedrockagentcore.EvaluatorInferenceConfig.html) — Optional inference configuration parameters that control model behavior during evaluation.

  **Default:** The foundation model's default inference parameters are used

  When not specified, the foundation model uses its own default values for
  maxTokens, temperature, and topP.

- `instructions`: string (required) — The evaluation instructions that guide the language model in assessing agent performance.

  These instructions define the evaluation criteria, context, and expected behavior.
  Instructions must contain placeholders appropriate for the evaluation level
  (e.g., `{context}`, `{available_tools}` for SESSION level).

  Note: Evaluators using reference-input placeholders (e.g., `{expected_tool_trajectory}`,
  `{assertions}`, `{expected_response}`) are only compatible with on-demand evaluation,
  not online evaluation.

- `modelId`: string (required) — The identifier of the Amazon Bedrock model to use for evaluation.

  Accepts standard model IDs (e.g., `'anthropic.claude-sonnet-4-6'`)
  and cross-region inference profile IDs with region prefixes
  (e.g., `'us.anthropic.claude-sonnet-4-6'`, `'eu.anthropic.claude-sonnet-4-6'`).

- `ratingScale`: [EvaluatorRatingScale](/cdk/api/v2/docs/aws-cdk-lib.aws_bedrockagentcore.EvaluatorRatingScale.html) (required) — The rating scale that defines how the evaluator should score agent performance.