nemoclaw-maintainer-evening

por nvidia

Executa a passagem de turno do mantenedor no final do dia para o NemoClaw. Verifica o progresso em relação à versão alvo, registra pendências para um incremento automático de pós-tag, gera uma passagem de QA…

npx skills add https://github.com/nvidia/nemoclaw --skill nemoclaw-maintainer-evening

NemoClaw Maintainer Evening

Close the day with one release candidate, one cumulative documentation change, and a clear handoff. Tagging is optional. Report tag creation first, then let the tag skill finish post-tag follow-through.

See PR-REVIEW-PRIORITIES.md for the daily cadence and the release-train policy for release rules.

1. Select the Target Version

Use the maintainer's exact vX.Y.Z when supplied. Otherwise, read the current target and show its merged and open work:

node --experimental-strip-types --no-warnings \
  .agents/skills/nemoclaw-maintainer-day/scripts/version-target.ts
node --experimental-strip-types --no-warnings \
  .agents/skills/nemoclaw-maintainer-day/scripts/version-progress.ts vX.Y.Z

Do not silently convert the answer into a patch, minor, or major bump. If nothing shipped, ask whether to stop without a tag.

2. Finish One Cumulative Documentation Change

Inspect the current Docs / Post-Merge Catch-Up state. The Pi workflow owns documentation catch-up for merged changes. Continue its managed draft PR when one exists. If no managed PR exists and the release entry is the only missing change, use one direct documentation-only PR.

The documentation PR must contain all required documentation for every merged change selected for the release and one canonical dated entry headed ## vX.Y.Z. Follow docs/CONTRIBUTING.md and obtain its required independent documentation writer review. Do not create a separate release-entry PR when the active cumulative docs PR can carry it.

Merge the documentation PR before selecting the tag candidate. A docs-only merge does not start another Docs / Post-Merge Catch-Up run. Preserve the merged PR, its final commit, its merge commit, and the final automated refresh coverage commit for the tag session.

If another product merge lands before candidate selection, decide whether it belongs in this release. When it does, update the cumulative documentation change first. When it does not, the tag skill may keep an earlier planned candidate that remains on main; later managed documentation work does not invalidate that candidate.

When an included merge changes the candidate after planning, generate a new exact-version plan after its documentation merges.

3. Show the Release Handoff

Show:

  • merged work in vX.Y.Z;
  • open PRs and issues still carrying vX.Y.Z;
  • the canonical release entry;
  • the cumulative docs PR, coverage commit, later commits and PRs, review state, and check state;
  • known image-publication state; and
  • the newest full E2E status, SHA, age, and URLs.

Open labeled items are post-tag planning state, not a tag blocker. State which items are expected to move to the next target, but do not perform label writes here.

4. Cut the Tag When Requested

Load nemoclaw-maintainer-cut-release-tag and pass the exact version. That skill owns:

  • the exact-version plan and candidate;
  • the required release entry, documentation coverage decision, and image evidence;
  • the maintainer's focused, full, or proceed E2E choice;
  • ../nemoclaw-release-vX.Y.Z/release-brief.md;
  • the exact confirmation phrase; and
  • signed tag creation and remote readback.

Do not run a full E2E suite automatically. Do not ask for confirmation until the release brief is complete and the maintainer has reviewed it.

5. Complete the Handoff

After tag readback, report:

  • tag and candidate commit;
  • plan and release-brief paths;
  • documentation coverage, maintainer decision, and image evidence URLs; and
  • E2E decision and exception reason, if any.

Then let nemoclaw-maintainer-cut-release-tag monitor the automatic post-tag workflows, draft the Announcement, and report lkg state. A post-tag failure does not change tag success.

Hard Rules

  • Never cut a tag without the maintainer's exact confirmation phrase.
  • Never bypass the release entry, documentation coverage decision, or applicable GHCR evidence.
  • Never make a different candidate stale merely because main advanced or a later documentation PR opened.
  • Never delay the tag-cut report for post-tag work. Continue the same task after that report.

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,…