Member Junction
    Preparing search index...

    An escalated record's answer from the target pipeline, already validated by it.

    interface EscalatedAnswer {
        AIPromptRunID?: string;
        ConstraintHash: string;
        EscalationReason: string;
        FeatureValueCacheID?: string;
        InputHash: string;
        Note: string;
        Payload: unknown;
        Pin: FeaturePipelinePinData;
        PipelineName: string;
        PromptID: string;
        PromptVersionHash?: string;
        SupersededNote: string;
        TargetOutputs: DataFeatureOutput[];
    }
    Index

    Properties

    AIPromptRunID?: string

    The target's MJ: AI Prompt Runs row (from its cache, when its answer was cached).

    ConstraintHash: string

    The target's constraint hash.

    EscalationReason: string

    Why the record escalated (Decision confidence below 0.7 (IsVIP 0.4)), for a failure message.

    FeatureValueCacheID?: string

    The target's own Dedup Cache entry for the answer, when the target caches.

    InputHash: string
    Note: string

    For the history row's reasoning: that the record escalated, to which pipeline, and from what confidence.

    Payload: unknown

    The target's validated payload, shaped by the target's output Refs.

    The target's pin the answer was computed under (it keys the answer's Feature Values).

    PipelineName: string

    The target pipeline's name, for a failure message.

    PromptID: string

    The MJ: AI Prompts ID of the target's prompt.

    PromptVersionHash?: string

    The target's prompt version hash for the answer.

    SupersededNote: string

    The reasoning of the history row that links the superseded decision's prompt run to the process run.

    TargetOutputs: DataFeatureOutput[]

    The target pipeline's outputs, so the payload can be read by output name.