# Kermt Setup

Bootstrap the KERMT agent environment, verify host docker + nvidia-container-toolkit...

- Repository: https://github.com/NVIDIA/skills (3812 stars)
- Category: Testing
- License: Apache-2.0
- Last updated: 2026-10-30
- Runs scripts: yes, review before use

## Install

Claude Code: `npx skills add NVIDIA/skills --skill kermt-setup -g -a claude-code -y`
Codex: `npx skills add NVIDIA/skills --skill kermt-setup -g -a codex -y`
Cursor: `npx skills add NVIDIA/skills --skill kermt-setup -g -a cursor -y`
Gemini CLI: `npx skills add NVIDIA/skills --skill kermt-setup -g -a gemini-cli -y`
Claude Code plugin marketplace: `/plugin marketplace add NVIDIA/skills`

## Description

Bootstrap the KERMT agent environment, verify host docker + nvidia-container-toolkit, build the kermt:latest image from the repo's Dockerfile if it doesn't yet exist, and run a GPU smoke test inside the container. Every other kermt-* skill depends on this; invoke it first.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# kermt-setup

Bootstrap the KERMT agent environment. Run this once on a fresh machine (or
after the Dockerfile or `environment.yml` changes) before invoking any other
`kermt-*` skill.

## Skill and runtime paths

Set `SKILL_DIR` to the absolute path of this installed skill directory. Export
`KERMT_REPO` as the absolute path to the KERMT checkout used for model
execution. The bundled container helper mounts that checkout at
`/workspace` and this skill at `/skill` (read-only). Commands inside
the container use `/skill/scripts/`.

## Hardware requirements

- **GPU**: at least one CUDA-capable NVIDIA GPU visible to the host. The image
  is based on `nvidia/cuda:12.6.3-cudnn-devel-ubuntu22.04`, so the host driver
  must support CUDA 12.6. Verify with host `nvidia-smi` before invoking.
- **Host docker**: docker engine + nvidia-container-toolkit. Without the
  toolkit, `docker run --gpus all` will fail at step 2 of the workflow below.
- **Disk**: ≈ 50 GB free for the built kermt image (`docker image inspect
  --format '{{.Size}}'` reports ≈ 44 GB; the `docker images` Size column
  can show ~100 GB because it counts shareable buildx attestation layers
  that are deduplicated across images). Plan for ~50 GB of unique on-disk
  storage; add a comfortable buffer if you're also keeping build cache.
- **Memory**: the build itself peaks at ~4 GB RAM during conda env solve.
- This skill does not run training/inference workloads itself; per-workflow
  hardware requirements (VRAM, GPU count) are declared in the respective
  `kermt-<workflow>` skills.

## When to invoke

- User explicitly asks (`/kermt-setup`, "set up kermt", "build the kermt image",
  etc.).
- Or another `kermt-*` skill detected that the image does not exist and routed
  here. (Most other skills call `kermt_ensure_image` themselves, so this is
  usually only needed for the first-time setup, debugging, or a forced rebuild.)

## Inputs

The rest: https://github.com/NVIDIA/skills/blob/HEAD/skills/bionemo-kermt-setup/SKILL.md