Skip to main content

Glossary

Agent​

Runnable logic that processes an input.

Dataset​

A collection of test entries, examples, or rules used during evaluation.

Dataset category​

An optional label on a dataset entry used to group related cases for later analysis.

Eval​

A named evaluation setup that connects agents and specifications.

Judge​

A scoring configuration used when correctness requires interpretation.

Organization​

The top-level shared workspace for members, projects, and collaboration.

Playground​

A faster exploratory run surface for trying configurations and generating derivative datasets.

Project​

The main container for assets and results.

Result​

A recorded output, score, or status produced during a run.

Red Team​

Evaluation work focused on misuse, harmfulness, jailbreaks, and failure-seeking behavior.

Run​

A single execution of an eval or playground workflow.

Secret​

A protected project-level value used by an agent at runtime.

Specification​

A reusable evaluation recipe that defines the datasets, attack methods, and related scoring setup for an eval.

Gold Team​

Evaluation work focused on desirable behavior, correctness, and robustness.