Skip to main content

Insert Dataset Rows

POST 
/datasets/:dataset_id/rows/insert

Insert new rows into a dataset's Iceberg table.

Each row is a dict of column values; _row_id is assigned automatically and must not be supplied. Applied as a single new Iceberg snapshot, so all rows in the request either land together or none do. When current_snapshot_id is supplied and it names a still-valid historical snapshot rather than the latest one, the table is rewritten from that base with the new rows appended instead of simply appending to the current head — an optimistic- concurrency guard against inserting against data the caller's view is already behind; an unresolvable snapshot ID is rejected as a conflict. The dataset's latest quality verdict is marked stale afterward, since the checked content has now changed.

Request​

Responses​

Successful Response