Member Junction
    Preparing search index...

    Interface DecisionDiscoveryOutcome

    What one run's decision discovery did. BaseAgent records it as the Agent discovery step.

    interface DecisionDiscoveryOutcome {
        Answer?: DecisionDiscoveryAnswer;
        CatalogSize?: number;
        DeclaredOptionCap?: number;
        HostAllowListSize?: number;
        Injected: boolean;
        Message?: string;
        NarrowedFrom?: number;
        OptionCount?: number;
        OptionLimit?: number;
        Reason?: string;
        Result?: AIDecisionRunResult;
        Succeeded: boolean;
        UncalibratedModel?: string;
        WithoutDescription?: string[];
    }
    Index

    Properties

    Answer? CatalogSize? DeclaredOptionCap? HostAllowListSize? Injected Message? NarrowedFrom? OptionCount? OptionLimit? Reason? Result? Succeeded UncalibratedModel? WithoutDescription?

    Properties

    Answer?: DecisionDiscoveryAnswer

    The answer, when there was a valid one.

    CatalogSize?: number

    How many agents the user may delegate to: the catalog before any narrowing.

    DeclaredOptionCap?: number

    The decision model's option cap, when one is declared.

    HostAllowListSize?: number

    How many IDs the host's allow-list held, when the run carried one. The catalog was cut to it.

    Injected: boolean

    Whether the <suggested_agent> message goes into the first prompt.

    Message?: string

    The message, when it is injected.

    NarrowedFrom?: number

    The number of options before the semantic search narrowed them. Set only when it did.

    OptionCount?: number

    How many options the Choice offered.

    OptionLimit?: number

    The most options the Choice may offer: see DecisionOptionLimit.

    Reason?: string

    Why nothing is injected.

    Result?: AIDecisionRunResult

    The decision call's result, for its usage.

    Succeeded: boolean

    False when discovery failed: an error, a timeout, a cancelled run, or an answer it could not use. True when the decision gave a valid answer, confident or not.

    UncalibratedModel?: string

    The model behind the answer (DescribeAnsweringModel), when it has no discovery calibration, so its answer was treated as unsure: discovery never suggests anything for it.

    WithoutDescription?: string[]

    The IDs of the catalog agents left out of the options for having no description.

    Settings

    Member Visibility

    On This Page

    Properties