The matching line plus a small window of the lines around it, joined by newlines and clipped to GREP_KNOWLEDGE_BASE_MAX_MATCH_LINE_CHARS. Wide spreadsheet rows are the normal case here, so the clip is what keeps a full result set from overflowing an agent's context window.
The id of the context the match was found in.
The source file name (or context title for text contexts).
The matching line, clipped to GREP_KNOWLEDGE_BASE_MAX_MATCH_LINE_CHARS with an ellipsis.
The 1-based line number the match starts on within its part.
The locator within the file: a sheet/section title when the extractor supplied one, else page N.
One matching line found by GrepKnowledgeBaseRequest.