Member Junction
    Preparing search index...

    Inputs to ConversationAgentRunner.processMessage. Replaces the original service's positional parameter list with a typed struct so call sites don't silently break when fields are added or reordered.

    interface ProcessMessageInput {
        AgentHistoryFrom?: Date | null;
        AllowedAgentIDs?: readonly string[] | null;
        appContext?: Record<string, unknown> | null;
        applicationId?: string | null;
        conversationDetailId: string;
        conversationId: string;
        explicitAgentId?: string | null;
        message: MJConversationDetailEntity;
        onProgress?: AgentExecutionProgressCallback;
        planMode?: boolean;
        requestedSkillIDs?: string[];
    }
    Index

    Properties

    AgentHistoryFrom?: Date | null

    The first moment of the conversation this turn may read. Sent to the server only when set; the server then loads the agent's history from there and skips its summary of earlier messages.

    AllowedAgentIDs?: readonly string[] | null

    The agents the conversation manager may delegate to. When set, ALL_AVAILABLE_AGENTS (the list the manager routes from) is narrowed to these. Null or omitted keeps every agent the user can run.

    appContext?: Record<string, unknown> | null

    Application context to inject into the agent's system prompt. Used by embedders (Form Builder cockpit, Component Studio AI Assistant, future LXP) to ground the agent in their UI state.

    applicationId?: string | null

    Application context for default-agent resolution. App-scoped Application Settings beat the global default.

    conversationDetailId: string

    The MJ: Conversation Details row that tracks this agent run on the server. Required for the optimized RunAIAgentFromConversationDetail path, which loads conversation history server-side.

    conversationId: string

    The conversation this message belongs to.

    explicitAgentId?: string | null

    Optional explicit agent ID. When set, wins over the DefaultAgentResolver chain. Mirrors the widget's [DefaultAgentId] input.

    The just-sent user message (used for latestMessageId plumbing).

    Optional progress callback. Mirrors the original service's signature — embedders forward this through to the widget's progress UI.

    planMode?: boolean

    Per-request Plan Mode toggle. When true (and the resolved agent's SupportsPlanMode capability is on), the root agent must present a plan for human approval before it may execute Actions/Sub-Agents. Defaults off — no behavior change unless set.

    requestedSkillIDs?: string[]

    Skill IDs the user requested via /skill-name mentions in the composer. Forwarded to the server, which intersects them with the agent's accepted skills AND the user's Run permission before any activate. Defaults to none.