Evaluating the following two trajectoriesEvaluation Metrics

tool_trajectory_avg_score

Evaluates function execution order, correct agent targets, and tool arguments.

expectedGatekeeper→Cameriere→Suggeritore
actualGatekeeper→Cameriere→Suggeritore
same steps, same order 1.0

response_match_score

Measures text similarity against target reference strings.

reference

"Here are 3 stops on your route with EV charging."

response

"I found 3 stops with EV chargers along your route."

similar, not identical 0.72