It would be useful to have (optional) metadata about the subdomains such as small descriptions (ta.wikipedia.org: "Tamil Wikipedia"), whether they use http or https by default, whether they redirect to a different URL (eg. alerts.google.com -> https://www.google.com/alerts), other more sensible things HN readers might come up with.
The reason I suggest this be put in a JSON/YAML/other-machine-plus-human-readable format is twofold: (a) to keep the original txt files simple and small (b) to not invent an ad-hoc format in the text files for the metadata, and force every user of the repo to write a parser (albeit a small one).
There's pros and cons to doing it as a single json file (with subdomains as keys) vs having a json/ folder with one json each corresponding to the txt files, but that's probably just a bikeshed point to barge through.
The idea behind this was to get the community to all chip in with website subdomains and then have 1 huge (freely available) list. This way it'd be easy to find all the subdomains for a particular domain if you want to do penetration testing or if you're just nosey like I am - can find some interesting stuff on random subdomains.
Maybe HN can help spread the word about it!