LLM Evaluation skill: what it does and how to install it

What it does. Establishes evaluation suites for LLM applications using automated metrics for generation, classification and retrieval, alongside human assessment and LLM-based judging. It can compare models or prompts, check outputs against references or without them, and track results over time.

When to use it. For measuring application quality, validating prompt changes, detecting regressions or investigating unexpected model behaviour.