chaos-engineering-resilience

Chaos engineering validates resilience by injecting controlled failures—network, compute, and I/O faults—into your distributed system and measuring recovery against steady-state metrics. This proves fault tolerance and disaster-recovery readiness before real outages occur.

Traditional testing checks known scenarios; chaos engineering proactively breaks systems in controlled ways across development, staging, and production to uncover unexpected weaknesses. It uses gradual blast-radius rollouts and automatic rollback to validate resilience safely.

Yes. This Skill applies chaos engineering to production with gradual blast-radius rollouts and automatic rollback triggers, letting you test real-world resilience while maintaining system stability through configurable failure scope and observable outcomes.

Define your target service, establish steady-state metrics (error rate, latency, throughput), select a fault type (network, CPU, disk, memory), set blast radius scope, and configure automatic rollback triggers. Agent-driven orchestration then runs and observes the experiment.

Measure baseline steady-state metrics—error rate, latency, throughput—before injecting faults, then monitor deviation from that baseline and blast radius during the experiment. This reveals how services respond to failures and whether recovery is automatic.

Runbooks document findings and automate recovery procedures discovered during chaos experiments. They codify what you learn about failures so teams can respond consistently when real incidents occur.