gpu-debian-setup

gpu-debian-setup is a skill for Claude Code, Codex from pedroiff0/awesome-skills. It costs 119 tokens per session (1,512 once invoked), scanned D, original, MIT.

A procedure for installing and checking NVIDIA proprietary drivers on Debian so local language-model software can use the GPU. It covers driver detection, Debian package settings, and verifying actual GPU memory use.

In plain words
What is it for?
Use it to configure NVIDIA GPUs for Ollama, llama.cpp, or vLLM on Debian, including Debian 13.
Why use it?
It addresses cases where local AI tools run on the CPU, `nvidia-smi` is missing, or Debian's driver packages do not work as expected.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/pedroiff0/awesome-skills/gpu-debian-setup
Any agent
npx skills add pedroiff0/awesome-skills --skill gpu-debian-setup
Clone the repo
git clone --depth 1 https://github.com/pedroiff0/awesome-skills

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for gpu-debian-setup

README.md
[![agentmods](https://agentmods.dev/badge/skills/pedroiff0/awesome-skills/gpu-debian-setup.svg)](https://agentmods.dev/skills/pedroiff0/awesome-skills/gpu-debian-setup)
Your own site
<a href="https://agentmods.dev/skills/pedroiff0/awesome-skills/gpu-debian-setup"><img src="https://agentmods.dev/badge/skills/pedroiff0/awesome-skills/gpu-debian-setup.svg" alt="Measured on agentmods" height="20"></a>
Per session 119 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,512 The whole file, excluding the scripts and references it only reads on demand.
Security scan D 3 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00119 $0.01512
Opus 5 $0.00060 $0.00756
Sonnet 5 $0.00024 $0.00302
Haiku 4.5 $0.00012 $0.00151

Measured 2d ago against content hash fcb23819bcb1, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade D, and why

gpu-debian-setup scanned grade D with 3 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

- Add `non-free` next to `non-free-firmware`, e.g. `sudo sed -i 's/ non-free-firmware/ non-free non-free-firmware/g' /etc/apt/sources.list.d/contrib.list`

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- Verify: `nvidia-smi` now lists the GPU. Cleanup: `rm -rf /tmp/nv_extract /tmp/NVIDIA.run`.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- `curl -fsSL -o /tmp/NVIDIA.run "https://us.download.nvidia.com/XFree86/Linux-x86_64/<VER>/NVIDIA-Linux-x86_64-<VER>.run"`
skills/mlops/gpu-debian-setup/SKILL.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

GPU driver setup on Debian for local LLM inference

When to use

  • User has an NVIDIA GPU but local models run slow / on CPU only.
  • "install nvidia driver", "enable GPU for ollama", "nvidia-smi not found", "ollama not using gpu".
  • After a driver install, nvidia-smi is missing even though the GPU seems present.

Steps (Debian, root or passwordless sudo)

  1. Assess hardware & current state (no root needed):
    • RAM/CPU: free -h, nproc, lscpu | grep -E "Model name"
    • GPU: lspci | grep -iE "vga|3d|nvidia"
    • Driver loaded? lsmod | grep -i nvidia ; nouveau still active? lsmod | grep -i nouveau
    • nvidia-smi (if present) ; CUDA: nvcc --version (often absent — normal)
    • Secure Boot: mokutil --sb-state (if available). OFF avoids module-signing pain.
    • DKMS headers: dpkg -l | grep linux-headers (needed to build the module).
  2. Enable the non-free component (Debian splits proprietary blobs out):
    • Find the active sources file (trixie commonly has /etc/apt/sources.list.d/contrib.list).
    • Add non-free next to non-free-firmware, e.g. sudo sed -i 's/ non-free-firmware/ non-free non-free-firmware/g' /etc/apt/sources.list.d/contrib.list
    • sudo apt-get update
  3. Install driver (metapackage pulls DKMS + userspace libs):
    • sudo apt-get install -y nvidia-driver firmware-misc-nonfree
    • trixie ships 610.x (CUDA repo) or 550.x (Debian non-free) — either works; don't mix versions.
  4. Blacklist nouveau (the open driver conflicts with the proprietary module):
    • echo -e 'blacklist nouveau\noptions nouveau modeset=0' | sudo tee /etc/modprobe.d/blacklist-nouveau.conf
    • sudo update-initramfs -u
  5. Reboot: sudo reboot. After reboot the proprietary module loads and nouveau is gone.

PITFALL — nvidia-smi is a DUMMY on Debian 13 (trixie)

On trixie, apt-get install nvidia-smi installs a transitional dummy package that ships NO binary. nvidia-utils does not exist by that name; neither nvidia-driver nor nvidia-driver-bin .debs contain /usr/bin/nvidia-smi; the CUDA-repo nvidia-smi .deb is also a dummy. Symptom: after a clean driver install + reboot, nvidia-smi -> "command not found", even though lsmod | grep nvidia shows the module loaded and libnvidia-ml.so.1 exists. FIX — extract the real binary from the official .run (must match the installed driver version!):

  • Version: dpkg -l | grep nvidia-driver (e.g. 610.43.02) or cat /proc/driver/nvidia/version.
  • curl -fsSL -o /tmp/NVIDIA.run "https://us.download.nvidia.com/XFree86/Linux-x86_64/<VER>/NVIDIA-Linux-x86_64-<VER>.run"
  • chmod +x /tmp/NVIDIA.run && /tmp/NVIDIA.run --extract-only --target /tmp/nv_extract
  • sudo cp /tmp/nv_extract/nvidia-smi /usr/bin/nvidia-smi && sudo chmod 755 /usr/bin/nvidia-smi
  • Verify: nvidia-smi now lists the GPU. Cleanup: rm -rf /tmp/nv_extract /tmp/NVIDIA.run. Important: the GPU works for compute WITHOUT nvidia-smi — it is only a monitoring tool. Do not block the whole setup on obtaining it. See references/nvidia-smi-extract.md.

Read the full file on GitHub · 66 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 66 lines · 119 tokens per session scan D fcb23819bcb1

Subscribe to this mod's changes

gpu-debian-setup is a skill published in the GitHub repository pedroiff0/awesome-skills (1 stars, last pushed 8d ago), licensed MIT. It adds 119 tokens to every session and 1,512 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it D with 3 findings (asks for root, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other skills, from other repositories

spark-environment-setup

Set up a working ML training/inference environment on NVIDIA DGX Spark (GB10, aarch64, CUDA 13). Use when installing PyTorch/Unsloth/TRL/vLLM on DGX Spark, hitting libcudart or wheel-ABI errors on aarch64, or choosing between NGC containers and bare pip installs.

wshobson/agents · 76 tokens

spark-memory-thermal-ops

Manage unified memory and thermals during long-running ML jobs on NVIDIA DGX Spark. Use when planning memory headroom for a training run on GB10, when a job OOMs on unified memory, or when monitoring temperature and power during multi-hour training.

wshobson/agents · 59 tokens

spark-training-gotchas

Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.

wshobson/agents · 63 tokens

local-inference-tuning

Select and tune a local LLM inference engine for the user's hardware. Use when setting up or auditing local/private model serving, choosing between MLX, llama.cpp, Ollama, and vLLM, estimating model fit, deciding cache/storage policy, tuning batching and KV cache flags, running smoke benchmarks, or exposing an…

markoblogo/abvx-agent-skills · 76 tokens

huggingface

Hugging Face Hub operations for robotics projects: inspect, download, create, upload, authenticate safely, explore Dataset Viewer data, run and diagnose Jobs, and inspect Spaces. Use when: "Hugging Face", "HF Hub", "hf download", "hf upload", "Hub dataset", "Hub model", "Dataset Viewer", "HF Jobs", or "Space logs" in…

robium-ai/robium · 129 tokens

ac-tools-video-query

Analyzes video files using Google Gemini API and answers questions about content. Triggers on keywords: video query, analyze video, video analysis, query video.

WaterplanAI/agentic-config · 35 tokens