observability
Observability - logging, metrics, tracing. Use when adding monitoring.
When & Why to Use This Skill
This Claude skill provides a comprehensive framework for system observability, focusing on the implementation of robust logging, metrics, and tracing. By integrating industry-standard tools like Sentry, PostHog, and Vercel, it enables developers to build transparent systems where errors are actionable, requests are traceable via end-to-end correlation IDs, and critical failures are caught instantly through optimized alerting.
Use Cases
- Case 1: Implementing end-to-end request tracing using correlation IDs to follow a single request through jobs and webhooks for rapid debugging.
- Case 2: Setting up real-time error tracking and actionable alerting with Sentry to ensure critical system failures are identified and resolved before impacting users.
- Case 3: Integrating PostHog analytics to eliminate monitoring blind spots and gain insights into system performance and user behavior.
- Case 4: Optimizing log quality and privacy by enforcing PII protection while ensuring logs contain sufficient context for 3 AM incident response.
- Case 5: Reducing alert fatigue by refining monitoring parameters to ensure only high-priority, actionable issues trigger notifications.
| name | observability |
|---|---|
| description | Observability - logging, metrics, tracing. Use when adding monitoring. |
Observability Guideline
Tech Stack
- Error Tracking: Sentry
- Analytics: PostHog
- Platform: Vercel
Non-Negotiables
- Correlation IDs must exist end-to-end (request → job → webhook)
- Alerts must exist for critical failures
- Errors must be actionable, not noise
Context
Observability is about answering questions when things go wrong. It's 3am, something is broken — can you figure out what happened? How fast?
PII protection in logs is enforced via privacy. This skill focuses on making debugging effective.
Driving Questions
- If something breaks now, how would we find out?
- What blind spots exist where errors go unnoticed?
- How long to trace a request through the system?
- What alerts exist, and do they fire correctly?
- Where is noise training people to ignore alerts?