Good question. I'm not sure. Probably we'll produce a complete read API first and then consider a write one. Maybe with proper credentialing there would be some value in it, since people have written lots of scraping clients that do it anyway.