debugging-observability
Installation
SKILL.md
Observability-Driven Debugging
Use this skill when
- Finding which dependency or path is slow
- Correlating errors to releases or deploys
- Using traces, metrics, and logs together
Do not use this skill when
- Code-only debugging without production data
- Hypothesis-free guessing
Instructions
- Start from the symptom: which SLO or user story broke.
- Correlate timing: deploys, traffic spikes, config changes.
- Use traces to find the slow hop; logs for the error context.
- Check segmentation: version, region, tenant, device.
- Rule out dependencies: latency to DB/cache/upstreams.
- Form a hypothesis with evidence; verify with next step.
- Document and synthesize findings to code actions.