CreateDecisionMetricParams

---

Interface: CreateDecisionMetricParams

Defined in: src/types/metrics.types.ts

Properties

answerOptions?

optional answerOptions:
  | string[]
  | DecisionLevel[];

Defined in: src/types/metrics.types.ts

Required for categorical (2-255 unique strings) and discrete (2-10 levels numbered 0, 1, 2… in order). Omit for boolean_probability.

---

description?

optional description: string;

Defined in: src/types/metrics.types.ts

---

groundTruth?

optional groundTruth: boolean;

Defined in: src/types/metrics.types.ts

---

inputType?

optional inputType: InputType;

Defined in: src/types/metrics.types.ts

---

modelName?

optional modelName: string;

Defined in: src/types/metrics.types.ts

Model registry version that answers the question. Defaults to jev-1.13.0.

---

name

name: string;

Defined in: src/types/metrics.types.ts

---

nodeLevel?

optional nodeLevel:
  | "agent"
  | "llm"
  | "retriever"
  | "tool"
  | "workflow"
  | "control"
  | "trace"
  | "session";

Defined in: src/types/metrics.types.ts

The node type the metric scores. Decision metrics score exactly one node type. Defaults to llm.

---

outputType?

optional outputType: OutputType;

Defined in: src/types/metrics.types.ts

One of boolean_probability, categorical or discrete. Defaults to the type that matches answerOptions: boolean_probability without options, categorical for strings, discrete for levels.

---

question

question: string;

Defined in: src/types/metrics.types.ts

The question the decision model answers (1-2,000 characters).

---

tags?

optional tags: string[];

Defined in: src/types/metrics.types.ts