observability-logging

Structured logging formats application events as JSON with consistent fields, making logs machine-readable and queryable. This Skill automates JSON-structured logs for AI/LLM pipelines, ensuring correlation IDs and LLM-specific metrics (tokens, cost, latency) are captured consistently across requests for end-to-end traceability.

OpenTelemetry tracing captures request flows across distributed systems using trace context and correlation IDs. For LLM apps, it enables visibility into token usage, inference latency, and cost per request, linking these metrics back to user actions and system components for observability.

LLM-specific metrics track tokens consumed, inference latency, and associated costs per API call. This Skill instruments your AI pipeline to automatically log these metrics in structured JSON format with correlation context, allowing you to analyze cost and performance by request, model, or user.

Yes. Correlation IDs propagate trace context across service boundaries, enabling end-to-end request tracking. This Skill applies correlation IDs automatically during LLM pipeline instrumentation, so logs and traces from different services remain linked for unified observability.

Structured logs use JSON with typed, consistent fields, enabling filtering, aggregation, and alerting by machines. Plain text logs require parsing and are error-prone. This Skill produces structured logs with redaction-friendly scaffolding, making sensitive data handling and compliance easier.