nvmini

name	nvmini
description	Codex built-in skill for lightweight NVIDIA GPU diagnostics and local CUDA visibility checks before running GPU-heavy experiments or benchmarks.
tags	["codex-builtin","nvidia","gpu","cuda","diagnostics"]

nvmini

When to Use

Use when the task needs quick GPU inventory, CUDA availability, memory pressure, driver visibility, or nvidia-smi style diagnostics.

Do Not Use When

Do not use for unrelated CPU-only Python debugging or broad system administration without GPU context.

Required Inputs

Local machine access and permission to run read-only GPU diagnostic commands.

Workflow

1. Check nvidia-smi or equivalent availability.
2. Record GPU model, memory, driver, CUDA visibility, and active processes.
3. Summarize whether the environment is ready for the requested workload.