runai-get-cluster-health-summary — Claude Code skill
runai-get-cluster-health-summary is a Claude Code skill that checks Run:ai cluster and node health, gives each pool a verdict, and identifies its main blocker. It shows why a pool won't schedule.
no license ★ 4,106 updated 2026-10-23
npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a claude-codeWhat runai-get-cluster-health-summary does
The skill takes a snapshot of cluster state via Run:ai MCP and delivers one verdict per pool plus a cluster-wide summary. It detects unready nodes, GPU allocation and free capacity, resource mismatches, pool saturation, and scheduler stalls—and tells the difference between a full pool and a broken scheduler. This is the entry point for ops triage.
Read-only by design: never restarts nodes, deletes pods, or changes quotas. Requires Run:ai MCP server with access to pools, node health, cluster inventory, and metrics. No Kubernetes fallback—without MCP, it stops.
Not for quota audits, GPU waste by project, single-workload diagnosis, or failure history. Related skills handle those: runai-audit-assigned-quota, runai-get-utilization-overview, and runai-diagnose-failed-workload.
What it helps with
- check whether a Run:ai cluster is healthy right now
- find out why a node pool stopped scheduling new workloads
- detect offline or failing GPU nodes
- see free GPU capacity at a glance
- distinguish a saturated pool from a stuck scheduler
Example prompts for the agent
- Show cluster health for
prod-us-west, breakdown by pool - Why isn't
gpu-a100scheduling? We have pending workloads piling up - How many GPUs are free? Which nodes are NotReady?
How to install runai-get-cluster-health-summary
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a claude-code2Manually: put the files in place
~/.claude/skills/runai-get-cluster-health-summary/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/runai-get-cluster-health-summary/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a codex2Manually: put the files in place
.agents/skills/runai-get-cluster-health-summary/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a cursor2Manually: put the files in place
.cursor/skills/runai-get-cluster-health-summary/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install NVIDIA/skills skills/runai-get-cluster-health-summary2Manually: put the files in place
.github/skills/runai-get-cluster-health-summary/SKILL.md1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a gemini-cli2Platform CLI
gemini skills install https://github.com/NVIDIA/skills --path skills/runai-get-cluster-health-summaryManually: put the files in place
.devin/skills/runai-get-cluster-health-summary/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about runai-get-cluster-health-summary
How do I install runai-get-cluster-health-summary in Claude Code?
Run npx skills add NVIDIA/skills --skill runai-get-cluster-health-summary -a claude-code, then open ~/.claude/skills/runai-get-cluster-health-summary/SKILL.md.
What does runai-get-cluster-health-summary do?
It snapshots your Run:ai cluster state: for each pool, it reports HEALTHY, DEGRADED, or STUCK, and names the blocker—unready nodes, failed GPUs, saturation, or scheduler stall.
Do I need kubectl or Kubernetes access?
No. The skill reads only through Run:ai MCP. If MCP is unavailable, it stops without falling back to kubectl.