What metadata should be added to a document loader to s...

A strong answer demonstrates an understanding of how to map natural language queries to structured metadata. Instead of simply listing fields, you must explain the provenance of the data. For instance, relying on an ‘ingested_at’ timestamp is often incorrect for temporal queries; the date must be extracted from the document content itself. Furthermore, you should be prepared to discuss which of these metadata fields can act as hard filters and why not all fields are suitable for such filtering.