AFAICS it's quite trivial implementation at the moment, and would otherwise require parsing the text file to find some references, right?
https://github.com/allegroai/clearml/blob/18c7dc70cefdd4ad739be3799bb3d284883f28b2/clearml/task.py#L1592