AgentSkillsCN

chaos-engineering-basics

在非生产环境及受控生产环境中，通过可控的故障注入来验证韧性假设的基础混沌工程工作流。当需设计具有明确爆炸半径控制的可控韧性实验时，可选用此流程；但请勿将其用于主动应急指挥或事后复盘文档的撰写。

SKILL.md
--- frontmatter
name: chaos-engineering-basics
description: "Foundational chaos engineering workflow for validating resilience assumptions through controlled fault injection in non-production and guarded production contexts. Use when designing controlled resilience experiments with explicit blast-radius controls; do not use for active incident command or postmortem document authoring."

Chaos Engineering Basics

Trigger Boundary

  • •Use when resilience assumptions need validation through controlled failure experiments.
  • •Do not use for incident postmortem documentation only; use incident-postmortem.
  • •Do not use when observability signals are not yet established; use observability-* first.

Goal

Improve reliability confidence by exposing hidden failure modes safely.

Inputs

  • •Service criticality and dependency map
  • •Existing SLO/SLI and alerting baseline
  • •Operational guardrails and blast-radius constraints

Outputs

  • •Chaos experiment plan with safety controls
  • •Hypothesis and expected steady-state metrics
  • •Findings, remediation actions, and re-test criteria

Workflow

  1. •Define steady-state behavior and safety guardrails.
  2. •Select one controlled fault scenario and blast radius.
  3. •Execute experiment with live monitoring and abort criteria.
  4. •Analyze impact against expected resilience behavior.
  5. •Publish remediation and schedule follow-up verification.

Quality Gates

  • •Experiment has explicit abort and rollback criteria.
  • •Observability is sufficient to detect degradation quickly.
  • •Blast radius remains within approved limits.
  • •Findings produce owned remediation actions.

Failure Handling

  • •Stop when guardrails or abort conditions are undefined.
  • •Escalate when experiment risk exceeds approved blast radius.