@squidcloud/client
    Preparing search index...

    Interface ExploreKnowledgeBaseGraphResponse

    Response structure for the bounded subgraph exploration (Phase 3 visualization).

    interface ExploreKnowledgeBaseGraphResponse {
        edges: KnowledgeBaseGraphEdge[];
        edgesTruncated?: boolean;
        nodes: KnowledgeBaseGraphNode[];
        scopeTruncated?: boolean;
        searchMatch?: "name" | "similar";
        topics?: KnowledgeBaseGraphTopic[];
        totalCount?: number;
    }
    Index

    The subgraph edges (outgoing only, both endpoints present in nodes). Capped: past the cap only the heaviest are returned and edgesTruncated is set.

    edgesTruncated?: boolean

    True when the nodes are linked by more relationships than edges carries.

    The subgraph nodes.

    scopeTruncated?: boolean

    True when contextMetadataFilter matched more contexts than one request reads; the subgraph then covers only the first of them.

    searchMatch?: "name" | "similar"

    How a search matched: by entity name, or by meaning when no name matched.

    The KB's topic (theme) nodes — the id→name/hierarchy table for the nodes' topicPath ids, so a visualization can label and group by theme without extra round trips. Present when the concept layer has been built (a rebuild ran). Under contextMetadataFilter, docCount counts only the matching contexts.

    totalCount?: number

    Size of the pool the subgraph was cut from — the entities in scope, the topic's members, the entity's neighbors or the search matches. Compare with nodes.length to offer more. Absent when not counted.