if we could do script.sh [<seeds>] and that pulled in only the relevant seeds for the developer, it could help skip the step of having to clone the entire test corpuses and/or having to manually use git sparse-checkout.