Netflix/chaosmonkey ? reverse-engineered prompt

Build me a Go based Chaos Monkey tool that integrates with Spinnaker and can randomly terminate running instances in a production environment to help test service resilience.

I want it to be able to read its settings from config, decide which app instances are eligible, and then pick targets at random on a schedule. It should support the kind of cloud backends Spinnaker works with, especially Kubernetes and common cloud providers, and it should be safe enough to use with clear checks before it kills anything. Include a small command line entry point, the core logic for scheduling and selecting instances, and the pieces needed to connect to Spinnaker and track what happens when outages are triggered.

Make it feel like a real production tool, with logs, error handling, and a straightforward way to deploy and run it. If you need to look up current docs online for Spinnaker or Chaos Engineering patterns, go ahead.