Skip to main content
skill CX Codex

runai-get-utilization-overview — Claude Code skill: GPU usage analysis

runai-get-utilization-overview is a Claude Code skill that checks whether projects and departments on a Run:ai cluster are using allocated GPUs or leaving them idle, and ranks them by utilization.

no license ★ 4,106 updated 2026-10-23

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add NVIDIA/skills --skill runai-get-utilization-overview -a claude-code
Open on GitHub

What runai-get-utilization-overview does

This skill answers one question: are allocated GPUs actually in use for computing, or sitting idle. It reads and reports only—never stops workloads, changes quota, or modifies pools.

It sorts GPU holders (projects, departments, node pools) into three categories based on average and p95 utilization: well-used, underutilized, or idle. You choose any time window; the default is seven days. When you drill into a project, the skill lists which workloads within it are consuming GPU capacity without real load.

Requires the Run:ai MCP server with access to GPU metrics and cluster inventory. If metrics are incomplete, a holder shows as UNKNOWN. It does not assess quota fairness, cluster health, individual workload status, or trends across windows.

What it helps with

  • identify which projects hold GPUs but aren't computing on them
  • measure how many GPU-hours go unused in a given period
  • rank node pools by wasted idle GPU capacity
  • list idle workloads within a specific project
  • check overall cluster utilization and find free capacity

Example prompts for the agent

  • Show GPU utilization by project for the last week on cluster-prod
  • Which department holds the most idle GPUs, and how many wasted hours over the past month?
  • List workloads in the ml-team project and tell me which ones are underutilized

How to install runai-get-utilization-overview

1Skills CLI (Vercel)

npx skills add NVIDIA/skills --skill runai-get-utilization-overview -a claude-code

2Manually: put the files in place

~/.claude/skills/runai-get-utilization-overview/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/runai-get-utilization-overview/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about runai-get-utilization-overview

How do I install runai-get-utilization-overview in Claude Code?

Run npx skills add NVIDIA/skills --skill runai-get-utilization-overview -a claude-code, then open ~/.claude/skills/runai-get-utilization-overview/SKILL.md.

What does runai-get-utilization-overview do?

It inspects GPU allocation and usage on your Run:ai cluster. It shows which projects, departments, or node pools hold GPU capacity that isn't actually being used, and quantifies the wasted hours by utilization tier.

Can this skill stop idle workloads?

No—it only reads. It will identify which workloads are idle and consuming GPU hours without load, but stopping them requires a separate write-capable workflow or manual action.

Similar skills

All skills for DevOps and infrastructure →

From the same set312

The whole set →
portfolio-optimization portfolio-optimization is a Claude Code skill for stock portfolio analysis with Mean-CVaR and Mean-Variance methods using NVIDIA cuOpt. Calculate allocations, assess risk, or backtest. CX Codex +1 ★ 4,106repo stars warp-eval warp-eval is a Claude Code skill that checks whether a specific hot code path is suitable for NVIDIA Warp, presenting measurements and evidence rather than an adoption verdict. CX Codex +1 ★ 4,106repo stars tao-run-deft-aoi-cosmos3 tao-run-deft-aoi-cosmos3 executes iterative training workflows for Cosmos3 visual inspection models. Designed strictly for binary OK or NG verification. CX Codex +1 ★ 4,106repo stars tao-run-deft-pas The tao-run-deft-pas skill runs repeated TAO CLIP or SigLIP training for image retrieval from labelled attributes. Use it when each training round should be evaluated against a stopping condition. CX Codex +1 ★ 4,106repo stars tao-run-on-docker tao-run-on-docker runs single-node NVIDIA TAO container tasks on a local Docker daemon or a remote GPU box over SSH without the TAO SDK. CX Codex +1 ★ 4,106repo stars rtvi-vlm-customize-model rtvi-vlm-customize-model guides VLM replacement in the VSS Alerts Blueprint and checks the three services that call the model separately. CX Codex +1 ★ 4,106repo stars tao-run-deft-cr-its-mining tao-run-deft-cr-its-mining runs an improvement loop for Cosmos Reason video classifiers, designed for binary tasks like collision detection from traffic cameras. CX Codex +1 ★ 4,106repo stars cdf-augmentation cdf-augmentation is a Claude Code skill that creates and validates Cosmos Data Factory YAML configurations and runs remote camera data augmentation pipelines. CX Codex +1 ★ 4,106repo stars

Project sponsors

Companies that keep this analytics open