What problem does it solve? When a service behaves differently across staging and production or a rollout's health is unclear, the root cause often lies in the platform layer rather than application code. This Skill provides a structured process to compare desired state (git) against running state (cluster), isolate the failing layer, and apply the smallest safe repair. ## Core Features & Use Cases - Desired vs Running State Comparison: Inspect live deployments with kubectl, helm diff, and revision checks to detect drift between source of truth and the cluster. - Layered Failure Isolation: Distinguish rollout orchestration, runtime config, secrets, networking, capacity, and application-code failures before patching anything. - Deep Diagnostics: Use ephemeral debug containers, eBPF tooling (Hubble, Tetragon), pprof profiling, and structured OpenTelemetry log correlation for in-depth investigation. - Use Case: A checkout service crash-loops only in production after a deploy. Use this Skill to check rollout status, compare the live image and env vars against git, inspect previous container logs, and roll back to the last known good state. ## Quick Start Use the debug-runtime-platform skill to investigate why the checkout service behaves differently in production than in staging after the latest rollout.
