Results and output

Online evaluation results are automatically saved to Amazon CloudWatch. By default, the service writes them as JSON log entries to a dedicated CloudWatch log group.

Log group structure

If you use the default dedicated destination, the service writes evaluation results to /aws/bedrock-agentcore/evaluations/results/<online-evaluation-config-id> . You can see the resolved result log group for a configuration on that evaluation’s details page in the Amazon Bedrock AgentCore console.

Each evaluation generates a separate log entry within this log group. Additionally, evaluation scores are emitted as CloudWatch metrics for monitoring and analysis.

Choose where results are written

Use resultDestination in outputConfig.cloudWatchConfig to choose where online evaluation results are written. If you don’t set it, the service uses a dedicated log group:

Use a custom output log group

When resultDestination is DEDICATED_LOG_GROUP, you can set logGroupName to send results to a log group that you choose. If the log group already exists, the service uses it as-is. If it doesn’t exist, the service creates it, so the evaluation execution role must allow logs:CreateLogGroup for that log group. The name can’t use the reserved /aws/bedrock-agentcore/evaluations/ prefix, except for this configuration’s own service-managed default group.

# Write results back to the trace source log group
outputConfig={"cloudWatchConfig": {"resultDestination": "SOURCE_LOG_GROUP"}}

# Write results to a custom dedicated log group
outputConfig={"cloudWatchConfig": {"resultDestination": "DEDICATED_LOG_GROUP", "logGroupName": "/my/team/evaluation-results"}}

Publish metrics to a custom namespace

Evaluation scores are also published as CloudWatch metrics. By default, the service uses the Bedrock-AgentCore/Evaluations namespace. To publish metrics under your own namespace, for example to separate results by team or tenant, set metricsNamespace in outputConfig.cloudWatchConfig. The value can’t begin with AWS/.

outputConfig={"cloudWatchConfig": {"metricsNamespace": "MyTeam/Evaluations"}}

Result format

Evaluation results follow OpenTelemetry semantic conventions for GenAI evaluation result events. When possible, each event is parented to the original span ID and includes the original trace ID and session ID.

You can use CloudWatch Logs Insights to query and analyze your evaluation results, and CloudWatch Metrics to monitor evaluation trends over time.

Viewing results in CloudWatch Observability Console

You can view and analyze your evaluation results using the CloudWatch Observability Console. The console provides visualizations, metrics, and detailed logs of your agent evaluations.

To view evaluation results

For more details, see AWS CloudWatch session trace evaluations documentation.

Viewing evaluation scores in CloudWatch Metrics

Evaluation scores are published as CloudWatch metrics. By default, they appear in the Bedrock-AgentCore/Evaluations namespace. If you set metricsNamespace, look for that custom namespace instead when you browse metrics in CloudWatch.

To view evaluation scores

For more details, see AWS CloudWatch session trace evaluations documentation.