Overview

Support CRUD operations and management of full datasets (input, output, context, etc.) in the Dev UI. This will add 1st class support for all evaluation use cases (eg: prod traces) in the Dev UI.

This is a blocker for agent evals and supporting interrupts in evals.

User goal(s)

Create and manage full datasets in the Dev UI

Requirements

Acceptance Criteria

• 1 Create full datasets
• 2 Edit, Update, Delete examples, delete dataset
• 3 Run evaluation (without inference) from the dataset