Chaos engineering для распределённых систем

testing

Chaos engineer is a Claude Code skill that designs chaos experiments, builds failure injection frameworks, and facilitates game day exercises for distributed systems — producing runbooks, experiment manifests, rollback procedures, and post-mortem templates. Reach for it when you want to know whether your service survives a pod or node failure in Kubernetes; when you need to find out if cascading latency to the database will take everything down; when you're planning a game day and need a scenario with explicit blast radius controls; or when you want chaos tests embedded in your CI/CD pipeline. The workflow covers architecture mapping and failure mode analysis, hypothesis and steady-state definition, monitored experiment execution, findings documentation, and automation. The skill works with Litmus Chaos (ChaosEngine manifests), Chaos Monkey, toxiproxy, Gremlin, and Pumba, and supports CI/CD integration. Built-in safety rules: verify baseline first, keep blast radius minimal and expand only after validation, automated rollback within 30 seconds. Use only on authorized environments with team awareness. The direction is injecting failures into live systems to improve resilience — not performance load testing (use dedicated load-testing tools for that).