From reading file_format.txt I gather that each header of each block has two bytes that could indicate the track and index number that the block belongs to. This could be used a bit like FLAC's CUESHEET metadata block, for computing FreeDB DiscIDs or extracting tracks from single file images, for instance.

If I understand correctly, what's missing today from the encoder is parsing cue sheets, and the guaranty that each track and each index starts precisely at the beginning of a block and ends precisely at the end of a block. Is that feature still planned for a future release?