What problem does it solve?
This skill addresses the gap between theoretical resilience patterns and actual system behavior by deliberately injecting faults to prove that circuit breakers, retries, and graceful degradation mechanisms function as intended under stress.
Core Features & Use Cases
- Hypothesis-Driven Testing: Enforces a rigorous four-part experiment structure (steady state, hypothesis, blast radius, rollback) to ensure chaos testing produces actionable data rather than anecdotes.
- Multi-Tier Fault Injection: Supports both in-process network fault injection via Toxiproxy and infrastructure-level faults like pod kills or network partitions using Chaos Mesh.
- Concurrency Validation: Includes specialized experiments to detect partial deadlocks and livelocks that standard Go runtime tools often miss.
Quick Start
Use the go-chaos-test skill to execute the pod-kill experiment against the canary environment to verify the system's automatic recovery capabilities.