Unfortunately, I don't think I can grab replies with the current system. Here's the JSON API url:
http://sabrina-dacos.tumblr.com/api/read/json?start=0&num=50
There's a bunch of nice 'photo-url-x' variables for the 'Summer Swirl' and 'Sprout' image posts, but the replies give the raw html, under the 'answer' variable:
{"id":"148746325376","url":"http:\/\/sabrina-dacos.tumblr.com\/post\/148746325376","url-with-slug":"http:\/\/sabrina-dacos.tumblr.com\/post\/148746325376\/are-you-very-sporty","type":"answer","date-gmt":"2016-08-10 17:10:50 GMT","date":"Wed, 10 Aug 2016 13:10:50","bookmarklet":0,"mobile":0,"feed-item":"","from-feed-id":0,"unix-timestamp":1470849050,"format":"html","reblog-key":"YvF8fFjk","slug":"are-you-very-sporty","question":"Are you very sporty?","answer":"<p>Just the occasional weight lifting\u2026<\/p><figure class=\"tmblr-full\" data-orig-height=\"870\" data-orig-width=\"600\"><img data-orig-height=\"870\" data-orig-width=\"600\" alt=\"image\" src=\"http:\/\/67.media.tumblr.com\/85581f6ebdaeba8c1a905ae2278b3d1e\/tumblr_inline_obpecqvICb1qdsnwr_540.jpg\"\/><\/figure>","tags":["female olympians","olympics","sporty girl","weight lifting","do you even lift bro"]}
I could parse the 'answer' variable for img tags, but then I'd hoover up any emoticons or other garbage. Maybe adding a test to make sure the data-orig-heigh/width product > x megapixels or something, but that sort of thing might ultimately be complicated and finicky in the current system, so I'd probably better be spending my time on the overhaul.
The thread watcher doesn't have that many default options, but you can set default check period and times to check under file->options->downloading. Also check the files and trash options page for by default, do not reimport files that have been previously deleted, which will apply to all new import contexts. There's no default tag options for the thread watcher–I'll make a note to add something.
Yeah, I'm overdue a help review. Some of the screenshots are very old. It is on my to-do to update everything, but I haven't got around to it yet. I'll make a note to add some better explanation as well. tl;dr about sync is that you download a whole load of json from my server and it ends up as tags on your files.
Adding a repo won't screw with your existing local tags–they get put in a completely different domain, and you can delete them any time without affecting anything you have already done. You can just go help->i don't know what I'm doing->just set up some repositories for me please if you like–it'll do all the key logistics for you.
Also if you are feeling hesitant about doing anything in hydrus, create/update your backup beforehand. If you have a decent backup somewhere, then you can screw around as much as you want, and rolling back to the functioning version will only take two minutes.
I solo dev this, and I am very happy to be free of the drama and other stress of group projects! I appreciate your enthusiasm, however. Feel free to look over the code–although be warned it isn't very high quality–and mess around with it if you like. Let me know if you have any questions about how it all goes together, or if you ever want advice on getting python and wx and sqlite to work together on your own projects.