Class ToolErrorEvaluator

java.lang.Object
dev.dokimos.core.BaseEvaluator
dev.dokimos.core.evaluators.agents.ToolErrorEvaluator
All Implemented Interfaces:
Evaluator

public class ToolErrorEvaluator extends BaseEvaluator
Detects tool execution failures by inspecting ToolCall.result().

This is a deterministic, glass-box evaluator (no LLM). A tool call is considered failed when any of the configured detectors fire:

The score is the fraction of tool calls that did not fail. An empty tool-call list scores 1.0, consistent with the other agent evaluators.
  • Method Details

    • builder

      public static ToolErrorEvaluator.Builder builder()
      Creates a new builder for constructing the evaluator.
      Returns:
      a new builder