output-dev-eval-testing

What problem does it solve?

Offline evaluation testing is required to validate AI workflows without impacting live production, enabling repeatable quality checks and regression detection for Output SDK pipelines.

Core Features & Use Cases

• Build, run, and manage offline eval tests for Output SDK workflows using @outputai/evals.
• Create dataset YAMLs, evaluators with verify(), and eval workflows to automate quality gates and reproducibility.
• Use for regression testing, subjective quality assessment with LLM judges, and CLI-driven validation across multiple scenarios.

Quick Start

Initialize an eval workflow scaffold and add datasets and evaluators to begin offline testing.