What problem does it solve? Diagnosing Linux server failures from collected evidence bundles, sosreports, and log archives is slow and error-prone, especially when host issues masquerade as Kubernetes or application problems. This Skill applies a structured, evidence-driven troubleshooting methodology to find root causes without jumping to conclusions. ## Core Features & Use Cases - Evidence Bundle Analysis: Ingests tarballs, sosreports, syslog/journalctl/dmesg logs, systemd state, and command outputs while preserving original artifacts as read-only. - Hypothesis-Driven Diagnosis: Maintains competing hypotheses with supporting and refuting evidence, timelines, and confidence levels across kernel, storage, network, resource, security, and container runtime domains. - Kubernetes Node Correlation: Correlates kubelet, containerd, CNI, and node lease evidence to determine whether host issues cause cluster symptoms. - Use Case: Given a sosreport from a node where pods keep evicting, the Skill identifies an OOM condition tied to a specific cgroup, builds the timeline, and cites exact log lines as evidence. ## Quick Start Use the linux-server-troubleshooting skill to analyze the evidence bundle in data/ and produce an evidence-backed root-cause analysis with a timeline and next actions.
