i4h-workflow-setup

작성자: nvidia

Verify host requirements and run `workflows/agentic/setup.sh`. Use when asked to set up, install, or bootstrap the agentic workflow, or hits missing `.venv`,…

npx skills add https://github.com/nvidia/skills --skill i4h-workflow-setup

i4h Workflow — Setup

Purpose

Verify host requirements and run the idempotent workflows/agentic/setup.sh to bootstrap the agentic workflow. Use when asked to set up, install, or bootstrap the workflow, or when hitting a missing .venv, third-party checkout, or engine error.

Base Code

These steps drive the i4h-workflows base code (the workflows/agentic/ tree). To reuse an existing checkout, set I4H_WORKFLOWS to its path (no clone happens). Otherwise this resolves the current repo, or clones to ~/i4h-workflows — pick that default without prompting. Run every command below from the resolved root:

# Resolve the i4h-workflows base code (provides workflows/agentic/).
ROOT="${I4H_WORKFLOWS:-$(git rev-parse --show-toplevel 2>/dev/null)}"
if [ ! -d "$ROOT/workflows/agentic" ]; then
  ROOT="${I4H_WORKFLOWS:-$HOME/i4h-workflows}"
  [ -d "$ROOT/workflows/agentic" ] || git clone https://github.com/isaac-for-healthcare/i4h-workflows "$ROOT"
fi
export I4H_WORKFLOWS="$ROOT"; cd "$ROOT"

Basics

  • workflows/agentic/setup.sh is the idempotent setup entry point.
  • Cosmos setup is separate; invoke only when the user asks for Cosmos or video transfer.

Preflight

command -v uv
command -v git
nvidia-smi
df -h .

Required: Linux, uv, git, NVIDIA driver/GPU, disk space for third-party checkouts. Docker is optional unless using Cosmos or local VLM containers.

Run the steps below in order. Each step is a separate bash call; variables persist in the local agent's tmux session.

Step 1 — resolve repo and run dir

REPO_ROOT="${I4H_WORKFLOWS:-$(git rev-parse --show-toplevel 2>/dev/null)}"; [ -d "$REPO_ROOT/workflows/agentic" ] || REPO_ROOT="$HOME/i4h-workflows"
RUN_DIR="${REPO_ROOT}/workflows/agentic/runs/setup_$(date +%Y%m%d_%H%M%S)"
mkdir -p "${RUN_DIR}/logs"
ln -sfn "${RUN_DIR}" "${REPO_ROOT}/workflows/agentic/runs/.latest"

Step 2 — run setup

On a freshly cloned repo the harness may ask to approve setup.sh once. Set I4H_WORKFLOWS to a pre-existing trusted clone to skip the fresh-clone gate.

"${REPO_ROOT}/workflows/agentic/setup.sh" > "${RUN_DIR}/logs/setup.log" 2>&1

Watch progress with tail -f "${RUN_DIR}/logs/setup.log". For component-specific retries (also idempotent):

"${REPO_ROOT}/workflows/agentic/third_party/setup.sh"
"${REPO_ROOT}/workflows/agentic/policy/gr00t_n16/setup.sh"

Verify

"${REPO_ROOT}/workflows/agentic/policy/run.sh" --list-envs
"${REPO_ROOT}/workflows/agentic/arena/run.sh" --list-envs
"${REPO_ROOT}/workflows/agentic/policy/run.sh" --env scissor_pick_and_place --dry-run
"${REPO_ROOT}/workflows/agentic/arena/run.sh" --env scissor_pick_and_place --dry-run

Prerequisites

  • Linux host with uv, git, and an NVIDIA driver/GPU (check via the Preflight commands).
  • Disk space for the third-party checkouts (df -h .).
  • Docker only when using Cosmos or local VLM containers; otherwise optional.

Limitations

  • Linux only; requires an NVIDIA driver/GPU.
  • Cosmos setup is separate — setup.sh does not run it; invoke Cosmos only when the user asks for it.
  • Verify steps use --list-envs and --dry-run; they confirm envs register and build, not that a full rollout works.

Troubleshooting

  • Error: uv / git / nvidia-smi not found — Cause: missing host requirement. Fix: install it and re-run Preflight before setup.sh.
  • Error: a single component fails partway through setup — Cause: a checkout or component venv did not finish. Fix: re-run that component's script (e.g. third_party/setup.sh, arena/setup.sh, policy/setup.sh); all are idempotent.
  • Error: --list-envs or --dry-run fails after setup — Cause: setup incomplete or a component venv missing. Fix: inspect ${RUN_DIR}/logs/setup.log (or workflows/agentic/runs/.latest/logs/setup.log) and re-run setup.sh or the failing component script.
  • Error: setup.sh fails instantly redirecting to a missing log dir — Cause: Step 2 ran before Step 1, so ${RUN_DIR}/logs does not exist. Fix: run Step 1 first, then Step 2.

Final Response

Report setup status, failed component (if any), relevant log path, next recommended smoke test.

nvidia의 다른 스킬

compileiq-debug
nvidia
무언가 잘못되었을 때 사용: Search()가 멈추거나, 모든 평가가 INVALID_SCORE를 반환하거나, 점수가 개선되지 않거나, 모든 설정이 동일한 숫자를 반환하거나, ptxas 오류 등이 발생할 때
create-github-pr
nvidia
gh CLI를 사용하여 GitHub 풀 리퀘스트를 생성합니다. 사용자가 새 PR을 만들거나, 코드 리뷰를 제출하거나, 풀 리퀘스트를 열고자 할 때 사용합니다. 트리거 키워드 -…
nemoclaw-maintainer-cross-issue-sweep
nvidia
다른 열린 이슈들을 스캔하여 주어진 PR이 함께 수정하거나 실수로 망가뜨릴 수 있는 이슈를 찾습니다. 인접 수정 기회와 모순 위험을 file:line…과 함께 출력합니다.
fhir-basics
nvidia
에이전트에게 FHIR R4 API의 작동 방식, 사용 가능한 리소스, 검색 매개변수를 사용한 쿼리 방법, 모든 응답 형식을 올바르게 파싱하는 방법을 가르칩니다…
compileiq-validate-result
nvidia
검색이 완료된 후, 속도 향상을 청구하거나 ACF를 발송하기 전에 사용합니다. dump_results CSV를 로드하고, 상위 K개 후보(단일 목표)를 추출합니다…
changelog-audit
nvidia
릴리스 전에 Warp CHANGELOG.md를 감사합니다: 누락된 항목 복구, 사용자 영향별 정렬, 항목 언어 다듬기, 줄 바꿈, (릴리스 브랜치 모드) 비교 업데이트…
maintain-dynamic-plugins
nvidia
NeMo Relay 동적 플러그인 로더, 매니페스트, Rust 네이티브 SDK, gRPC 워커 프로토콜, Python 워커 SDK, 문서, 테스트 및 릴리스 워크플로 커버리지를 유지 관리합니다.
dgx-diagnose
nvidia
일반적인 DGX Station GB300 문제 진단 — CUDA 충돌, 잘못된 GPU 타겟팅, vLLM/SGLang 컨테이너 버그, MIG 상태 문제, NVLink/Fabric Manager 오류,…