AgentSkillsCN

prod-debug-kit

借助可重复使用的调试工具包，对生产环境中的突发事件进行分类与稳定处理。在出现错误、延迟激增、内存泄漏或间歇性故障时使用，同时兼顾快速遏制与根源分析的需求。

SKILL.md
--- frontmatter
name: prod-debug-kit
description: Triage and stabilize production incidents with a repeatable debugging kit. Use when errors latency spikes memory leaks or intermittent failures appear, and when quick containment plus root-cause analysis are both required.

Prod Debug Kit

Workflow

  1. •Stabilize first. Define blast radius and immediate containment steps. Prefer reversible mitigations such as feature flags, rate limiting, or temporary fail-open paths.
  2. •Build a minimal repro. Correlate logs, metrics, and traces by request ID and time window. Reduce to the smallest reproducible path.
  3. •Isolate root cause. Check dependency health, resource saturation, retries timeouts, and recent deploy changes. Form and test one hypothesis at a time.
  4. •Implement durable fix. Apply minimal-risk code or config change. Add guards such as circuit breaker, bounded retries, or idempotency protections.
  5. •Verify and prevent recurrence. Confirm error rate and latency recovery. Add alert tuning and targeted regression tests.

Output Contract

Always provide:

  • •Incident timeline with concrete timestamps.
  • •Containment action and business impact.
  • •Root cause with evidence links.
  • •Follow-up prevention tasks.