review-github-pr

par nvidia

Examiner une demande de tirage GitHub en résumant son diff et les décisions de conception clés. À utiliser lorsque l’utilisateur souhaite examiner une PR, comprendre les modifications dans une branche, ou obtenir un…

npx skills add https://github.com/nvidia/openshell --skill review-github-pr

Review GitHub Pull Request

Summarize a GitHub pull request diff, highlighting key design decisions and notable code snippets.

Prerequisites

  • The gh CLI must be authenticated (gh auth status)
  • You must be in a git repository with a GitHub remote

Step 1: Resolve the PR

The user will provide either a PR number (e.g., #123 or 123) or a branch name. Determine which input was given and resolve it to a PR.

If a PR number is provided

Strip any leading # and use the numeric ID directly. Proceed to Step 2.

If a branch name is provided

Look up the open PR whose head branch matches:

gh pr list --head "<branch>" --state open
  • If exactly one PR is found, extract its number and proceed to Step 2.
  • If multiple PRs are found, list them and ask the user which one to review.
  • If no PR is found, skip Step 2 (no PR description to fetch) and go directly to Step 3 using the local git diff fallback.

Step 2: Fetch PR Description

Retrieve the PR metadata:

gh pr view <number> --json title,body,state,headRefName,baseRefName,labels,author

Record the title, body, headRefName, and baseRefName for use in later steps.

Step 3: Generate the Diff

Primary: gh pr diff

Fetch the diff via the GitHub CLI:

gh pr diff <number>

If this succeeds, use this diff and proceed to Step 4.

Fallback: local git diff

If no PR exists (branch-only case) or the gh diff command fails, fall back to a local diff:

# Ensure both branches are available locally
git fetch origin <target-branch> <source-branch>

# Generate the diff
git diff origin/<target-branch>...origin/<source-branch>

If the user provided a branch name and no PR was found, diff against main:

git fetch origin main <branch>
git diff origin/main...origin/<branch>

Handling large diffs

If the diff output is very large (thousands of lines), use the Task tool to process it in chunks. Summarize each chunk independently, then merge the summaries. Do not skip or truncate parts of the diff — accuracy depends on reading all of it.

Step 4: Analyze and Summarize

Read through the full diff (and the PR description if available). Produce a summary with the following sections. Keep every section as concise as possible — brevity is a priority.

Summary format

## PR Review: <title>

**PR:** [#<number>](<url>)  <- only if a PR exists
**Author:** <author>
**Branch:** `<source>` -> `<target>`

### Overview
<1-3 sentences describing what this PR does and why>

### Key Design Decisions
- <decision 1 with file:line reference>
- <decision 2 with file:line reference>
- ...

### Notable Code
<short fenced code snippets that illustrate the most important changes -- max 3 snippets>

### Potential Concerns  <- omit if none
- **<concise user-visible behavior>** — Before this PR, <affected persona>
  experienced <previous behavior>. With this PR, <new concerning behavior>, so
  <user-visible impact>. Details: `<file>:<line>`.

Guidelines for the summary:

  • Overview: State what changed and why. Pull context from the PR description if available.
  • Key Design Decisions: Focus on why something was done a particular way, not what changed. Include file_path:line_number references. Examples: choice of algorithm, new abstraction introduced, API contract change, migration strategy.
  • Notable Code: Include only the most instructive or surprising snippets. Keep each snippet under 15 lines. Always include the file path above the code block.
  • Potential Concerns: Only include genuine risks that warrant a change or a deliberate accept/reject decision. Describe each concern in terms of observable behavior for the affected persona, such as a sandbox creator, sandbox user, operator, administrator, SDK consumer, or developer maintaining the system. Always compare the previous behavior with the new concerning behavior and state the resulting user-visible impact. Prefer the compact form: "Before this PR, <persona> experienced <old behavior>. With this PR, <new behavior>, so <impact>." Add only the minimum file and line references needed to substantiate the finding.
    • Use the PR base as the normal previous-behavior baseline. Review older history only when the change is fixing or extending an earlier feature and that history is necessary to explain the behavioral contract. In that case, describe the relevant transitions explicitly: "Before <commit>, ... After <commit>, ... With this PR, ...".
    • Translate internal failure modes and race conditions into what the affected person would observe. Internal implementation details belong in the trailing file and line references, not in place of the behavior description.
    • Do not assign P0/P1/P2 or similar priority labels. The behavioral comparison and impact should give maintainers enough context to accept or reject the suggested change.
    • Do not fabricate concerns or claim a behavioral regression without evidence for both the prior and proposed behavior.
  • Agent infrastructure: When the PR changes behavior, commands, or development workflows, use the sync-agent-infra maintenance map to check that related skills were updated. When it adds, removes, or renames skills or crates; changes workflow relationships or skill coverage; modifies issue or PR templates; or changes agent cross-references, apply the full consistency checklist. Report missing companion updates or drift under Potential Concerns.

Step 5: Output

Print the summary directly in the chat as formatted markdown.

If the user requests it, also save the summary to a file:

# Default path
reviews/<number>-review.md

# Or for branch-only reviews
reviews/<branch-name>-review.md

Useful Commands Reference

CommandDescription
gh pr list --head <branch>Find PR by head branch
gh pr diff <number>Get PR diff
gh pr view <number> --json ...Get full PR metadata
git diff origin/<target>...origin/<source>Local diff between branches

Example Usage

Review by PR number

User says: "Review PR #456"

  1. Fetch PR metadata for number 456
  2. Fetch diff via gh pr diff 456
  3. Produce summary

Review by branch name

User says: "Review branch feature/add-pagination"

  1. Look up PR with gh pr list --head "feature/add-pagination"
  2. If found, fetch PR metadata and diff
  3. If not found, diff against main locally
  4. Produce summary

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,…