Netra Simulation Setup

Use this skill to design realistic multi-turn simulation datasets and evaluate conversational agent behavior.

When To Use

• You need to test multi-turn behavior, not only single-turn outputs.
• You want to compare agent performance across user personas.
• You need pre-production validation for goal achievement and factual correctness.

Simulation Building Blocks

• Evaluators: Session-level scoring of conversation quality and outcomes.
• Multi-turn datasets: scenario name, scenario, persona, max turns, user data, and facts.
• Test runs: Conversation transcript + evaluator results + trace links.