All public data from xd is available in three .zip archives:
xd-metadata.zip (2MB), metadata and similarity for all puzzles in the corpus, public and private
xd-puzzles.zip (12MB), all 5,688 puzzles in the corpus published before 1965, in .xd format: 5,633 from the New York Times (1942-1964), 50 from the first Simon & Schuster crossword books (1924), and a handful of early newspaper puzzles
xd-clues.zip (67MB), over 6,000,000 answer/clue usages, grouped by publication-year
Email crosswords for comparison or inclusion to submitxd@saul.pw as attachments of .puz, .ipuz, .xd, or a .zip of any of these. Each filename should include the date (in YYYYMMDD format), or a three digit number (NNN) if they are published sequentially.
If you have more than a few puzzles, please email me at xd@saul.pw to discuss the best way to import and organize them.