@squidcloud/client
    Preparing search index...

    Interface KnowledgeBaseGraphBuildUsage

    LLM usage of one graph build (rebuildGraph run), kept per build so a later build cannot erase it. Accumulated across the run's execution attempts — a resumed rebuild adds to the same record.

    interface KnowledgeBaseGraphBuildUsage {
        buildId: string;
        completedAt?: number;
        contextsReextracted: number;
        mode: KnowledgeBaseGraphRebuildMode | "resolution";
        stages: KnowledgeBaseGraphBuildStageUsage;
        startedAt: number;
        status: "in_progress" | "completed" | "failed";
        usage: KnowledgeBaseGraphUsage;
    }
    Index
    buildId: string

    Identity of the build, stable across its execution attempts.

    completedAt?: number

    When the build reached a terminal state (epoch millis); absent while it runs.

    contextsReextracted: number

    Contexts whose chunks the build re-extracted; 0 when it re-extracted nothing.

    mode: KnowledgeBaseGraphRebuildMode | "resolution"

    The plan the build executed; 'resolution' is the internal resolution-only continuation.

    The split of usage across the stages the build ran.

    startedAt: number

    When the build started (epoch millis).

    status: "in_progress" | "completed" | "failed"

    The build's lifecycle state.

    The build's total LLM usage, with its per-model split.