prepare-pr

par nvidia

Préparer, ouvrir, créer, publier, mettre à jour ou modifier une pull request NeMo Relay ou un corps de PR avec les bons tests, la documentation, l'hygiène du contributeur et le pull du dépôt…

npx skills add https://github.com/nvidia/nemo-relay --skill prepare-pr

Prepare A PR For NeMo Relay

Companion Guidance

Use karpathy-guidelines alongside this skill for implementation or review work. Keep changes scoped, surface assumptions, and define focused validation before editing.

Use this skill at the end of a contributor or maintainer change before opening a pull request. Also use it whenever a user asks to create, open, publish, update, or edit a NeMo Relay pull request, pull request description, or PR body.

If this repo-local guidance conflicts with generic GitHub publishing, connector, or plugin guidance, this skill wins for PR body format, validation language, and review handoff details.

Checklist

  • Branch scope is coherent and reviewable
  • Relevant tests passed under validate-change
  • Changed files were formatted with the language-native formatter
  • Any Rust change ran just test-rust
  • Any Rust change ran cargo fmt --all
  • Any Rust change ran cargo clippy --workspace --all-targets -- -D warnings
  • crates/core or crates/adaptive changes ran the full language matrix
  • Targeted uv run pre-commit run --files <changed files...> checks were used during iteration where useful
  • uv run pre-commit run --all-files passed or issues are understood
  • Docs and examples updated for any public behavior changes
  • Dependent maintainer or consumer skills updated when code changes affected their APIs, bindings, commands, paths, packaging guidance, or best practices
  • Pull request title follows Conventional Commit style and uses the correct type
  • Pull request body follows .github/pull_request_template.md
  • Breaking changes or renamed surfaces are called out explicitly

Pull Request Title

Use Conventional Commit style for PR titles:

<type>: <concise imperative summary>

Choose the type from the actual change surface, not from the impact of the review comment or CI outcome. Use fix only for an actual user-facing or runtime/product code bug fix. Never use fix for changes that are not related to product code behavior, including chores, CI configuration, docs, tests, packaging metadata, generated-output handling, or agent/skill guidance.

Common examples:

  • ci: update codecov coverage reporting
  • docs: clarify release workflow
  • enhancement: improve scope configuration
  • chore: refresh generated attribution data
  • skills: add a plugin-authoring guide
  • test: add Python scope regression coverage
  • fix: preserve scope-local middleware cleanup

Opening A Pull Request

Always use .github/pull_request_template.md as the source of truth for the PR body. Before opening a PR, read the current template and preserve its headings, checkboxes, comments' intent, and related-issue guidance.

This applies both when creating a new PR and when editing an existing PR description. Do not use a generic Summary / Why / Validation body unless the current repository template uses those headings.

When using GitHub CLI, prefer:

gh pr create --template .github/pull_request_template.md

If a tool cannot consume the template directly, create the PR body from the template content and then fill in every visible section before opening the PR. Do not replace the template with a freeform summary.

After creating or editing a PR, fetch the rendered PR body and verify that the template's visible headings and checklist items are still present.

The PR body must include:

  • #### Overview with a concise summary and both contribution confirmation checklist items preserved
  • #### Details with the concrete changes made
  • #### Where should the reviewer start? with the most useful file, test, or design decision
  • #### Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to) with an issue reference, or a clear Relates to: none entry when there is no related issue

Only check the contribution confirmation boxes when they are true. If either confirmation cannot be made, stop before opening the PR and surface the blocker.

References

  • CONTRIBUTING.md
  • .github/pull_request_template.md
  • validate-change

Plus de skills de nvidia

compileiq-debug
nvidia
Utilisez quand quelque chose ne va pas : Search() bloque, toutes les évaluations retournent INVALID_SCORE, les scores ne s'améliorent pas, chaque configuration retourne le même nombre, erreurs ptxas…
create-github-pr
nvidia
Créer des pull requests GitHub en utilisant l'interface en ligne de commande gh. Utiliser lorsque l'utilisateur souhaite créer une nouvelle PR, soumettre du code pour révision, ou ouvrir une pull request. Mots-clés de déclenchement -…
nemoclaw-maintainer-cross-issue-sweep
nvidia
Analyse les autres problèmes ouverts pour trouver ceux qu’une PR donnée pourrait également corriger ou casser accidentellement. Génère des opportunités de correctifs adjacents et des risques de contradiction avec fichier:ligne…
fhir-basics
nvidia
Apprend aux agents comment fonctionnent les API FHIR R4, quelles ressources sont disponibles, comment les interroger avec des paramètres de recherche, et comment analyser correctement tous les formats de réponse…
compileiq-validate-result
nvidia
Utiliser APRÈS qu'une recherche soit terminée et AVANT de réclamer un accélérateur ou d'expédier un ACF. Charge le CSV dump_results, extrait les K meilleurs candidats (mono-objectif)…
changelog-audit
nvidia
Auditer le CHANGELOG.md de Warp avant une publication : récupérer les entrées perdues, trier par impact utilisateur, affiner le langage des entrées, ajuster les retours à la ligne et (en mode branche de publication) mettre à jour la comparaison…
maintain-dynamic-plugins
nvidia
Maintenir les chargeurs de plugins dynamiques NeMo Relay, les manifestes, les SDK natifs Rust, le protocole worker gRPC, le SDK worker Python, la documentation, les tests et la couverture du workflow de publication
dgx-diagnose
nvidia
Diagnostiquer les problèmes courants du DGX Station GB300 — plantages CUDA, ciblage incorrect du GPU, bugs de conteneur vLLM/SGLang, problèmes d'état MIG, erreurs NVLink/Fabric Manager,…