Member Junction
    Preparing search index...

    Interface DiscoveryEvalEnvironment

    What one discovery run needs from the environment, loaded the way production's discovery loads it: the agents the run's user may run, the conversation manager that runs discovery, the decision prompt's option cap, and the semantic search.

    interface DiscoveryEvalEnvironment {
        AllAgents: readonly { ID: string; Name: string | null }[];
        ConversationManager: { ID: string; Name: string | null };
        DeclaredCap?: number;
        Prompt: MJAIPromptEntityExtended | null;
        RunnableAgents: readonly DecisionDiscoveryAgent[];
        SearchFor?: (request: string) => DecisionDiscoverySearch;
    }
    Index

    Properties

    AllAgents ConversationManager DeclaredCap? Prompt RunnableAgents SearchFor?

    Properties

    AllAgents: readonly { ID: string; Name: string | null }[]

    Every agent in the engine's catalog: what an

    may name.

    ConversationManager: { ID: string; Name: string | null }

    The conversation manager (Sage): never an option, and the agent whose

    does not count.

    DeclaredCap?: number

    The decision prompt's option cap (DecisionPromptOptionCap), when one is declared.

    Prompt: MJAIPromptEntityExtended | null

    The decision prompt, from AIEngine. Null for a baseline cell, which makes no call.

    RunnableAgents: readonly DecisionDiscoveryAgent[]

    The agents the run's user may run (DecisionDiscoveryRunnableAgents).

    SearchFor?: (request: string) => DecisionDiscoverySearch

    The semantic search for a request (DecisionDiscoveryAgentSearch), or undefined when the provider cannot run it.

    Settings

    Member Visibility

    On This Page

    Properties