You can clone with
HTTPS or Subversion.
We need a way to standardize input data (possibly for different areas: social sciences, sciences, humanities, law), and easily visualize diffs.
Let me take a crack at what kinds of data we need. Let's consider a few categories, based on area. Feel free to suggest additions, as I'm not really sure after the generic.
For Sciences, a patent.
And we basically need a bookmarklet, hooked up to citeproc-js, that will create a visual preview of this:
The github API includes pull requests, each of which includes a link to the raw diff file; so perhaps that could provide an easy way to do it.