Pipeline Evaluator

Purpose

Score a completed agent pipeline across 5 dimensions and produce a structured evaluation report with a composite verdict (EXCELLENT/GOOD/ACCEPTABLE/NEEDS_IMPROVEMENT) and ranked recommendations.

When to Invoke

Skill({ skill: 'pipeline-evaluator' })

Invoke when:

• A multi-task pipeline has completed and quality measurement is needed
• Router requests post-pipeline quality gate
• Reflection agent needs structured metrics for the rubric
• Continuous improvement loop requires baseline data