It is an inventive use of Github.  I definitely like directories of personal sites.  Personal site directories are best populated by webmaster URL submissions, because it’s hard for an editor to figure out the themes of a site in a timely manner, and one problem here is that there is a lot of friction to submitting a site, although the email workaround is thoughtful.

I can’t help but wonder about building a similar directory site that aggregates its data by Webmention and uses the h-cards from websites to automatically update itself.

Hrrm.  What happens if the directory admin finds a simply stunning site that has no webmention or h-card capability?  Do we abdicate our human judgement and automatically disqualify a more than worthy site because it does not have a barcode we like?  I always have a problem with using just code up front as a criteria for inclusion.  Or perhaps my bar is set too low: 1. it has a web address, preferably it’s own domain, 2. the site renders and is readable to humans, 3. the site has good content.  I do think a directory such as you describe is worth trying as a long term experiment.The strength of a directory over a search engine is the human editing. Search engines cannot measure quality, only popularity.  Why give up, even partially, one of your few advantages?The second thing, I don’t think a directory in 2019 can rely on webmaster submissions alone to grow the directory.  A whole new generation of webmasters have come of age without directories, webrings and the like.  They don’t understand the need to submit (silos), they don’t know they should submit their URL, they don’t know how to submit URL’s, they don’t know how a directory works or how to search one.  I think any new directory has to just offer URL submission, but in the end go out and find good sites.