Read the listing response completely
Check whether the response includes a next-page token, cursor, or truncation indicator. The absence of a file in the first page does not prove that the object is missing from the bucket.
Use a narrow prefix or filter to reduce the result set when possible, but make sure the prefix boundary is correct. Similar keys can be included or excluded unexpectedly if the delimiter or character matching is misunderstood.
Process pages without skipping entries
Use the continuation value returned by one page to request the next. Avoid changing the prefix, sort assumptions, or filters midway through iteration, because that can create gaps or duplicates.
If the listing is used for deletion or migration, first collect or review the exact affected keys and keep a record of progress. Test on a small prefix before operating over an entire bucket, and confirm the final count against the intended scope.