> Also worth noting we'll be storing "greatest hits" copies of the ~15,000 most-starred / most-relied-on repos, along with a sampling of several thousand repos

> Also worth noting we'll be storing "greatest hits" copies of the ~15,000 most-starred / most-relied-on repos, along with a sampling of several thousand repos with few/no stars Making all of this code essentially useless. You'd need to store those repos and their entire dependency tree.