Test-Driven Development asserts we write tests for our application first, and then write code that adheres to that specification. This allows us freedom when refactoring, but it also gives us a clear signal when we have the minimum amount of code to pass that test. How do we borrow this pattern for our infrastructure-as-code? Configuration management offers some capability here, and Kubernetes sets a nice pattern with readiness and liveness probes, but we can go further. In this talk I'll introduce an additional pattern that demonstrates how we can write our monitoring checks first, and get real world results. I'll also cover a basic implementation of this pattern, showing how to put it into practice in your organization.