Skip to content

KubeMD v0.1.0 — Evidence-first Kubernetes diagnosis with case memory

Latest
Latest

Choose a tag to compare

View all tags
@guiyi-labs guiyi-labs released this 16 Aug 11:35
· 6 commits to main since this release
v0.1.0
5022d3f

KubeMD v0.1.0

Evidence-first runtime diagnosis for Kubernetes failures — with case memory. A DSH (DeepSeek Harness) skill that diagnoses live, broken clusters, not manifests.

What ships

  • SKILL.md — 7-phase diagnosis procedure (feedback loop → signals → falsifiable hypotheses → dry-run fix → case record)
  • 5 verified playbooks: CrashLoopBackOff / OOMKilled / Network / Pending / NodeNotReady
  • signal-map.md symptom→signal→command cheatsheet
  • Scripts: collect-signals.sh (one-shot signal collection), record-case.sh (case library append)
  • cases.yaml example library + full verification report (real fault-injected kind cluster, 5 scenarios, no hard errors)

Install (30s)

git clone https://github.com/guiyi-labs/kubemd ~/.dsh/skills/dsh-k8s-diagnosis

CLI twin

The same deterministic engine ships as a go install CLI: aiops diagnose / aiops cases (https://github.com/guiyi-labs/aiops-platform, v0.1.0 binaries available).

Verification report

Assets 2
Loading