i4h-workflow-e2e

por nvidia

Execute o pipeline agentic completo de ponta a ponta (record → mimic → annotate → replay → convert → visualize → finetune → validate). Use quando solicitado a executar todo o…

npx skills add https://github.com/nvidia/skills --skill i4h-workflow-e2e

i4h Workflow — End-to-End

Purpose

Run the full end-to-end agentic pipeline (record, mimic, annotate/filter, replay, convert, visualize, finetune, validate). Use when the user asks to run the full pipeline, smoke the whole workflow, demo the workflow, or do an e2e run.

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

  • Env config (source of truth): workflows/agentic/config/environments/<env>.yaml — drives every stage for <env> (robot, task, policy, cameras, arena.max_timesteps, dataset.* mappings).
  • Use the e2e script for full pipeline runs.
  • For per-stage work, use the corresponding dataset/finetune/validate skills.
  • assemble_trocar is inference-only; the e2e script skips finetune and checkpoint validation for it.

Dry Run

REPO_ROOT="${I4H_WORKFLOWS:-$(git rev-parse --show-toplevel 2>/dev/null)}"; [ -d "$REPO_ROOT/workflows/agentic" ] || REPO_ROOT="$HOME/i4h-workflows"
"${REPO_ROOT}/workflows/agentic/scripts/e2e/run.sh" --dry-run --env <env>

Run

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

For Claude Code --print or any other noninteractive runner, keep Step 2 in the foreground. This is a validation requirement: do not use Claude background tasks, async task mode, Bash background mode, &, nohup, tmux, disown, or any detached process/task id, and do not answer that the pipeline is still running. Do not return until run.sh exits and you have inspected logs/SUMMARY.txt on success, or the failing stage log on failure. Report the run dir, skipped stages, per-stage status, key artifacts, and cleanup/stop status before finishing.

Step 1 — setup

REPO_ROOT="${I4H_WORKFLOWS:-$(git rev-parse --show-toplevel 2>/dev/null)}"; [ -d "$REPO_ROOT/workflows/agentic" ] || REPO_ROOT="$HOME/i4h-workflows"

Step 2 — e2e pipeline

"${REPO_ROOT}/workflows/agentic/scripts/e2e/run.sh" --env <env>

Flags

  • --skip-mimic, --skip-annotate, --skip-replay, --skip-viz
  • --from-stage <stage> --run-dir <existing-run> resumes from a prior run.
  • Policy record/verify stages open the sim window by default. Set ARENA_HEADLESS=1 before run.sh only when the user explicitly asks for headless/no-window execution.

Stages: setup record mimic annotate replay convert viz finetune validate summary.

Outputs

The script prints RUN_DIR and symlinks it to runs/.latest. Subdirs:

  • logs/ — per-stage logs, workflow.log (full teed output), and logs/SUMMARY.txt (the final summary report)
  • data/
  • lerobot/
  • checkpoint/ (trainable envs only)

Monitor

run.sh runs every stage in the foreground and returns only when the whole pipeline ends, so track a long run from a separate shell (do not expect to query it from the shell that launched it):

tail -f "${REPO_ROOT}/workflows/agentic/runs/.latest/logs/workflow.log"   # live per-stage progress
cat    "${REPO_ROOT}/workflows/agentic/runs/.latest/logs/SUMMARY.txt"     # final report (once DONE)

Stop

Step 3 — stop (if needed)

"${REPO_ROOT}/workflows/agentic/stop.sh" all --env <env>

Prerequisites

  • Workflow set up via [[i4h-workflow-setup]] (the .venv must exist); setup is also the first pipeline stage.
  • A valid --env name to drive the run.
  • For per-stage work, use the corresponding dataset/finetune/validate skills instead.

Limitations

  • assemble_trocar is inference-only; the e2e script skips finetune and checkpoint validation for it.
  • checkpoint/ outputs are produced for trainable envs only.
  • Resuming requires both --from-stage <stage> and --run-dir <existing-run>.

Troubleshooting

  • Error: .venv not found / module import fails - Cause: workflow not set up. Fix: run [[i4h-workflow-setup]] first.
  • Error: env not recognized - Cause: wrong --env name. Fix: pass a valid env name; dry-run first with --dry-run --env <env>.
  • Error: resume fails to find prior outputs - Cause: --from-stage used without a matching --run-dir. Fix: pass --from-stage <stage> --run-dir <existing-run>.
  • Error: stale processes block a rerun - Cause: a previous pipeline session is still running. Fix: run stop.sh all --env <env> before retrying.

Final Response

Report env, run dir, skipped stages, per-stage success/failure, key artifact paths.

Mais skills de nvidia

compileiq-debug
nvidia
Use quando algo está errado: Search() trava, todas as avaliações retornam INVALID_SCORE, as pontuações não estão melhorando, toda configuração retorna o mesmo número, erros de ptxas…
create-github-pr
nvidia
Crie pull requests do GitHub usando a CLI gh. Use quando o usuário quiser criar um novo PR, enviar código para revisão ou abrir um pull request. Palavras-chave de acionamento -…
nemoclaw-maintainer-cross-issue-sweep
nvidia
Escaneia outras issues abertas para encontrar aquelas que um determinado PR pode também corrigir ou quebrar acidentalmente. Gera oportunidades de correção adjacentes e riscos de contradição com arquivo:linha…
fhir-basics
nvidia
Ensina aos agentes como funcionam as APIs FHIR R4, quais recursos estão disponíveis, como consultá-los com parâmetros de busca e como analisar corretamente todos os formatos de resposta…
compileiq-validate-result
nvidia
Use APÓS a conclusão de uma Pesquisa e ANTES de reivindicar qualquer aceleração ou enviar um ACF. Carrega o CSV dump_results, extrai os K melhores candidatos (objetivo único)…
changelog-audit
nvidia
Auditar o CHANGELOG.md do Warp antes de um lançamento: recuperar entradas perdidas, ordenar por impacto ao usuário, refinar a linguagem das entradas, ajustar quebras de linha e (no modo de branch de lançamento) incrementar comparação…
maintain-dynamic-plugins
nvidia
Manter carregadores de plugins dinâmicos do NeMo Relay, manifestos, SDKs nativos em Rust, protocolo de worker gRPC, SDK de worker Python, documentação, testes e cobertura do fluxo de lançamento
dgx-diagnose
nvidia
Diagnostique problemas comuns do DGX Station GB300 — falhas de CUDA, direcionamento incorreto de GPU, bugs de contêiner vLLM/SGLang, problemas de estado MIG, erros de NVLink/Fabric Manager,…