tsuga-analyze-trace-latency

What problem does it solve?

This Skill helps you identify which operations are driving a service’s slow requests by analyzing trace latency across a chosen time window and separating transient spikes from sustained degradation.

Core Features & Use Cases

• Operation-level latency ranking: Computes p95 latency grouped by span.name and highlights the slowest operations.
• Sustained vs transient diagnosis: Determines whether latency remains elevated across consecutive 5-minute windows.
• Evidence-based correlation with errors: Searches logs for ERROR events during the worst latency window to support (or refute) error-related causes, and can optionally correlate via trace_id when both traces and logs exist.

Quick Start

Use tsuga-analyze-trace-latency for service checkout-api over the last 2 hours to find which operations have the worst p95 latency and whether it is sustained.