Package akka.javasdk.agent.evaluator

package akka.javasdk.agent.evaluator

Built-in LLM-as-judge agents (ToxicityEvaluator, SummarizationEvaluator, HallucinationEvaluator) that call a model to judge a piece of text or an answer against a reference.

Related Packages

Package	Description
akka.javasdk.agent	Agent components, plus the guardrails and classifiers that govern their model and tool calls.
akka.javasdk.agent.autonomous	AutonomousAgent components — agents that run their own iteration loop against a task queue — with their definition, state, and runtime notifications.
akka.javasdk.agent.task	Typed, durable tasks for autonomous agents: Task and TaskTemplate definitions, the TaskEntity that records each task's lifecycle, TaskRule result validation, and the events and notifications the entity publishes.

Class	Description
HallucinationEvaluator	An agent that acts as an LLM judge to evaluate whether an output contains information not available in the reference text given an input question.
HallucinationEvaluator.EvaluationRequest	An answer to judge for hallucinated content.
HallucinationEvaluator.Result	The verdict for an answer: whether it passed the hallucination check, and the judge's explanation for the label it assigned.
SummarizationEvaluator	An agent that acts as an LLM judge to evaluate a summarization task.
SummarizationEvaluator.EvaluationRequest	A summary to judge against the document it was produced from.
SummarizationEvaluator.Result	The verdict for a summarization task: whether the summary passed the quality check, and the judge's explanation for the label it assigned.
ToxicityEvaluator	An agent that acts as an LLM judge to evaluate if an AI response or other text is racist, biased, or toxic.
ToxicityEvaluator.Result	The verdict for a piece of text: whether it passed the toxicity check, and the judge's explanation for the label it assigned.