Consider running ecosystem-analyzer on all projects (including failing ones)

[sharkdp]

This let's us get rid of good.txt and bad.txt. And we would get notified if we "accidentally" fix a panic in one of the remaining bad.txt projects.