Class RubricEngine

Index

Constructors

Accessors

Methods

Agreement Consensus ConsensusForSubject CreateDraft Diagnostics Evaluate EvaluateRecord GetRubric StartHumanEvaluation SubjectContent

Constructors

constructor

new RubricEngine(
    evaluations?: RubricEvaluationStore,
    records?: RubricRecords,
    services?: RubricEvaluatorServices,
): RubricEngine
Parameters
- evaluations: RubricEvaluationStore = unsetEvaluations
- records: RubricRecords = emptyRecords
- services: RubricEvaluatorServices = {}
  What the engine lends every evaluator: the prompt and decision services and the agent runner. A call can replace any of them through EvaluateParams.services.
Returns RubricEngine
- Defined in Rubrics/Engine/src/RubricEngine.ts:154

Accessors

`Static`Instance

get Instance(): RubricEngine
The process-wide engine, backed by the Global Object Store so bundler duplication across execution paths cannot split it. Constructing another engine does not rebind this one.

Returns RubricEngine
- Defined in Rubrics/Engine/src/RubricEngine.ts:139

Methods

Agreement

Agreement(ratings: number[][], minimumSample?: number): AgreementResult
Kappa and alpha, withheld below the sample floor.
Parameters
- ratings: number[][]
- OptionalminimumSample: number
Returns AgreementResult
- Defined in Rubrics/Engine/src/RubricEngine.ts:353

Consensus

Consensus(
    scores: number[],
    method?: "Mean" | "Median" | "TrimmedMean",
    trim?: number,
): ConsensusResult
Mean, median, or trimmed mean of normalized scores, with spread.
Parameters
- scores: number[]
- Optionalmethod: "Mean" | "Median" | "TrimmedMean"
- Optionaltrim: number
Returns ConsensusResult
- Defined in Rubrics/Engine/src/RubricEngine.ts:348

ConsensusForSubject

ConsensusForSubject(
    input: {
        contextEntityId?: string;
        contextEntityName?: string;
        contextRecordId?: string;
        major?: number;
        method?: "Mean" | "Median" | "TrimmedMean";
        rubricId?: string;
        rubricName?: string;
        subjectEntityId?: string;
        subjectEntityName?: string;
        subjectRecordId: string;
    },
): Promise<ConsensusResult>
Loads the subject's Submitted scores and returns the consensus. The caller does not pass the scores.
Parameters
- input: {
      contextEntityId?: string;
      contextEntityName?: string;
      contextRecordId?: string;
      major?: number;
      method?: "Mean" | "Median" | "TrimmedMean";
      rubricId?: string;
      rubricName?: string;
      subjectEntityId?: string;
      subjectEntityName?: string;
      subjectRecordId: string;
  }
  - OptionalcontextEntityId?: string
  - OptionalcontextEntityName?: string
  - OptionalcontextRecordId?: string
  - Optionalmajor?: number
    When omitted, the latest Published major. Scores from other majors are left out.
  - Optionalmethod?: "Mean" | "Median" | "TrimmedMean"
  - OptionalrubricId?: string
  - OptionalrubricName?: string
  - OptionalsubjectEntityId?: string
  - OptionalsubjectEntityName?: string
  - subjectRecordId: string
Returns Promise<ConsensusResult>
- Defined in Rubrics/Engine/src/RubricEngine.ts:303

CreateDraft

CreateDraft(
    input: {
        nodes: RubricNodeSnapshot[];
        rubricId?: string;
        rubricName?: string;
    },
): Promise<{ id: string; status: "Draft" }>
Stores the payload as a Draft version. Throws when the stored status is anything else. Publishing stays a human action.
Parameters
- input: { nodes: RubricNodeSnapshot[]; rubricId?: string; rubricName?: string }
Returns Promise<{ id: string; status: "Draft" }>
- Defined in Rubrics/Engine/src/RubricEngine.ts:341

Diagnostics

Diagnostics(
criteria: { key: string; notApplicable: number; scores: number[] }[],
): DiagnosticFlag[]
Item-analysis flags for a published major version's criteria.
Parameters
- criteria: { key: string; notApplicable: number; scores: number[] }[]
Returns DiagnosticFlag[]
- Defined in Rubrics/Engine/src/RubricEngine.ts:358

Evaluate

Evaluate(
params: EvaluateParams,
): Promise<
{ evaluation: RubricEvaluationRecord; output?: RubricEvaluatorRun },
>
Creates the named evaluator, runs it, saves the draft, and submits it. A throw from the evaluator becomes a Failed evaluation carrying the message. An unregistered name, or an evaluator a person completes such as Human, throws before anything is saved.
Parameters
- params: EvaluateParams
Returns Promise<{ evaluation: RubricEvaluationRecord; output?: RubricEvaluatorRun }>
- Defined in Rubrics/Engine/src/RubricEngine.ts:165

EvaluateRecord

EvaluateRecord(input: EvaluateRecordInput): Promise<EvaluateRecordResult>
Resolves the rubric name or id to its latest Published version, then calls evaluate. This is not the Get Rubric action.
Parameters
- input: EvaluateRecordInput
Returns Promise<EvaluateRecordResult>
- Defined in Rubrics/Engine/src/RubricEngine.ts:233

GetRubric

GetRubric(
input: { rubricId?: string; rubricName?: string; versionId?: string },
): Promise<RubricVersionSnapshot>
The version tree for a name, id, or a specific version. Does not score. A version id returns that version. Otherwise this is the latest Published version.
Parameters
- input: { rubricId?: string; rubricName?: string; versionId?: string }
Returns Promise<RubricVersionSnapshot>
- Defined in Rubrics/Engine/src/RubricEngine.ts:290

StartHumanEvaluation

StartHumanEvaluation(
    input: {
        assigneeId: string;
        drafts: EvaluationDraftStore;
        rubricId: string;
        rubricName: string;
        subjectEntityId: string;
        subjectRecordId: string;
        tasks: RubricTaskStore;
        versionId: string;
    },
): Promise<{ evaluationId: string; taskId: string }>
Creates a Draft and a task for the assignee. Does not score. The human fills the draft in and submits it later.
Parameters
- input: {
      assigneeId: string;
      drafts: EvaluationDraftStore;
      rubricId: string;
      rubricName: string;
      subjectEntityId: string;
      subjectRecordId: string;
      tasks: RubricTaskStore;
      versionId: string;
  }
Returns Promise<{ evaluationId: string; taskId: string }>
- Defined in Rubrics/Engine/src/RubricEngine.ts:196

SubjectContent

SubjectContent(
input: { subjectEntityName: string; subjectRecordId: string },
): Promise<RubricSubjectContent>
The content an evaluator may read for one subject record. Read-only. Does not score and does not publish.
Parameters
- input: { subjectEntityName: string; subjectRecordId: string }
Returns Promise<RubricSubjectContent>
- Defined in Rubrics/Engine/src/RubricEngine.ts:429

Class RubricEngine

Index

Constructors

Accessors

Methods

Constructors

constructor

Parameters

Returns RubricEngine

Accessors

StaticInstance

Returns RubricEngine

Methods

Agreement

Parameters

Returns AgreementResult

Consensus

Parameters

Returns ConsensusResult

ConsensusForSubject

Parameters

OptionalcontextEntityId?: string

OptionalcontextEntityName?: string

OptionalcontextRecordId?: string

Optionalmajor?: number

Optionalmethod?: "Mean" | "Median" | "TrimmedMean"

OptionalrubricId?: string

OptionalrubricName?: string

OptionalsubjectEntityId?: string

OptionalsubjectEntityName?: string

subjectRecordId: string

Returns Promise<ConsensusResult>

CreateDraft

Parameters

Returns Promise<{ id: string; status: "Draft" }>

Diagnostics

Parameters

Returns DiagnosticFlag[]

Evaluate

Parameters

Returns Promise<{ evaluation: RubricEvaluationRecord; output?: RubricEvaluatorRun }>

EvaluateRecord

Parameters

Returns Promise<EvaluateRecordResult>

GetRubric

Parameters

Returns Promise<RubricVersionSnapshot>

StartHumanEvaluation

Parameters

Returns Promise<{ evaluationId: string; taskId: string }>

SubjectContent

Parameters

Returns Promise<RubricSubjectContent>

Settings

On This Page

`Static`Instance

`Optional`contextEntityId?: string

`Optional`contextEntityName?: string

`Optional`contextRecordId?: string

`Optional`major?: number

`Optional`method?: "Mean" | "Median" | "TrimmedMean"

`Optional`rubricId?: string

`Optional`rubricName?: string

`Optional`subjectEntityId?: string

`Optional`subjectEntityName?: string