Puppet autosigning should send a chill up any security-minded Operations engineer. It does for me. If you can't trust the Puppet Agent, then you can't trust the security of anything *deployed* by Puppet. In this session you will watch Yelp's journey from using naive autosigning (and not trusting anything from Puppet) to a real autosigning policy that we can trust. Once in place, we can move all of the poorly-implemented alternative security and secret mechanisms into something sane: Puppet. Once you can trust Puppet again, you don't have to fight your tools, and you can instead build on top of them!