Design end-to-end chaos engineering experiments as complete, runnable manifests. Use this skill whenever the user says "create a chaos experiment", "inject a fault", "design a chaos test", "I want to test resilience of X", "add a chaos step to my pipeline", or needs to produce ChaosExperiment / ChaosEngine YAML for LitmusChaos or Harness Chaos Engineering. Also trigger when the user wants to test failure scenarios such as pod crashes, network partitions, disk exhaustion, or CPU spikes, even if they don't use the word "chaos".
