What problem does it solve? When VMs become unhealthy, Terraform state drifts, or CI tests flake repeatedly, engineers waste hours hunting through logs and system state. This Skill performs read-only diagnostics across KVM hosts, VMs, Terraform state, and CI artifacts, then returns an evidence-backed root-cause hypothesis and routes the fix to the right specialist agent. ## Core Features & Use Cases - Read-only diagnostics: Collects logs (journalctl, docker logs), system state (pf vm check, virsh list, terraform plan), and metrics (df, free, ss) without mutating anything. - Known-pattern matching: Maps symptoms to a catalog of documented failure patterns such as Terraform state drift, non-idempotent Ansible roles, Nexus 503/404 mirror issues, etcd timeouts, and SSH agent timeouts. - Root-cause delegation: Routes each diagnosed issue to the appropriate agent (/platform-engineer, /devops, /ops-engineer, /qa-engineer) with a structured report including priority. - Use Case: A k8s-matrix CI run fails three times in a row on the same environment. The Skill analyzes past QA artifacts, identifies a common apt upstream timeout pattern, and delegates a Nexus mirror diagnosis plus an Ansible timeout fix. ## Quick Start Invoke the reliability-engineer skill with a description of the symptom, for example: diagnose why VM k8s-ha-alt-adp2 shows kubelet not ready after reapply.
