@squidcloud/client
    Preparing search index...

    Interface GrepKnowledgeBaseResponse

    Response structure for a literal scan over a knowledge base's raw extracted text.

    interface GrepKnowledgeBaseResponse {
        limitation?: GrepScanLimitation;
        matches: AiKnowledgeBaseGrepMatch[];
        truncated: boolean;
        truncatedContentFileNames?: string[];
        unscannedContextCount?: number;
    }
    Index
    limitation?: GrepScanLimitation

    Why the scan did not cover the full requested scope, if it did not. Absent means it did — which is the only case where an empty matches is evidence that the pattern does not occur. The whole point of this API is to make absence claims trustworthy, so callers must branch on this before concluding anything from an empty result.

    The matches found, in context/part/line order.

    truncated: boolean

    True when the returned list stopped at maxMatches — more matches exist than were returned. This means ONLY that: every other way a scan can fall short of the request lives in limitation, so a caller branching on this one field never mistakes an incomplete scan for "there are more hits".

    truncatedContentFileNames?: string[]

    Every in-scope document known to have lost content at ingest, by file name. Present whenever such a document is in scope, alongside whichever limitation the scan reported — a cut document is no less cut because the scan also stopped early. Absent when none is.

    unscannedContextCount?: number

    With limitation: 'partialCoverage', how many contexts in scope have no stored text and were therefore never scanned. The result is silent about those contexts.