assess-quality

To evaluate LLM output quality, you can use a structured mechanism that assesses outputs against task requirements, producing a numerical score and rationale to decide whether further iteration is needed.

You can automate prompt iteration by applying a quality threshold to LLM outputs. The evaluation produces an XML verdict with a 0.0-1.0 score, which can be used in pipelines to decide if a result is production-ready.

Assessing LLM correctness and completeness involves evaluating outputs against four criteria: Correctness, Completeness, Clarity, and Quality. This rubric ensures a structured score and rationale for the generated text.

Yes, automated quality scoring applies to code generation, content creation, and data-to-text tasks. It evaluates the LLM output against your specified task and outputs a structured XML verdict suitable for automation.

The best way to run an LLM quality assessment via CLI is by executing a command with your task and output parameters, which generates an XML file containing the score and verdict to review if further iterations are required.