Skip to main content

`POST /v1/internal/files/debatch`

POST 

/v1/internal/files/debatch

The chunk count derives from the file's size_bytes (the CHUNK_RECORDS-record batch shell per chunk); record parsing rides the instruction-file-spec version registered for the file's class.

Execute the debatch_file step: split the file's records into chunked instruction batches (design §10). The step is idempotent on file_id:chunk_index - a workflow retry re-enters here and the terminal-file guard makes completed work a no-op.

Responses​

OK