Skills github-pr-review

📦

github-pr-review

Name: github-pr-review
Author: Bae-ChangHyun

Content revision r1 High Risk ⚙️ External commands📁 Filesystem access

Résoudre les commentaires de revue de PR GitHub

Les retours de revue peuvent être dispersés entre fichiers, niveaux de sévérité et threads. Cette compétence organise les commentaires, guide les corrections, vérifie les changements et publie des réponses claires.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Maintain this Skill? Claim it ⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "github-pr-review" from https://skillstore.io/skills/bae-changhyun-github-pr-review.md and its manifest at https://skillstore.io/api/skills/bae-changhyun-github-pr-review/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/bae-changhyun-github-pr-review.md Signed manifest GET /api/skills/bae-changhyun-github-pr-review/manifest Signed lockfile GET /api/skills/bae-changhyun-github-pr-review/lockfile

Test it

Using "github-pr-review". Résous tous les commentaires de haute sévérité sur ma pull request actuelle.

Expected outcome:

Résumé des sévérités avec les commentaires critiques et élevés en premier.
Un plan de correction proposé pour chaque commentaire bloquant.
Une checklist de vérification avant les étapes de commit et de réponse.

Using "github-pr-review". Aide-moi à répondre aux commentaires de revue après les avoir corrigés.

Expected outcome:

Texte de réponse court pour chaque thread résolu.
Notes pour les commentaires reportés ou intentionnels.
Un rappel pour confirmer avant de publier les réponses.

Using "github-pr-review". Prépare les commits pour les retours de revue sur cette PR.

Expected outcome:

Regroupement de commits suggéré par changements fonctionnels et cosmétiques.
Sujets de messages de commit alignés avec les recommandations de la compétence.
Une checklist finale pour les tests, le linting et les conversations non résolues.

Security Audit

High Risk

v8 • 7/6/2026 Open versioned report

The skill is a GitHub PR review workflow that intentionally relies on GitHub CLI, git, and project test commands. Several static findings are Markdown false positives, but real command execution, repository side effects, prompt-injection wording, and untrusted PR comment handling remain material risks.

Files scanned

200

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (3)

High

SKILL.md:3

Prompt Injection Attempt Detected

The skill frontmatter says "MUST use this skill" and "OVERRIDES default behavior." This attempts to elevate the skill above normal agent decision rules.

The suspicious wording is explicit in the skill description and uses override language. This matches a direct prompt-injection pattern against higher-priority instructions.

High

SKILL.md:20-23 SKILL.md:64-78

Untrusted Review Comments Can Drive Agent Actions

The workflow fetches PR comments and tells the agent to analyze, fix, verify, commit, and reply to each comment. It does not warn that review comments are untrusted input.

The cited lines make PR comments the driver for code changes and thread replies. Malicious review text could steer the agent unless handled as untrusted input.

High

SKILL.md:126-142 SKILL.md:183-190

Automated Push and Review Actions

The skill instructs agents to push commits and submit formal PR reviews, including approval examples. These actions can affect repository state and merge decisions.

The cited workflow includes git push, gh pr review commands, and an always-submit-review rule. These are high-impact actions if run without strong human authorization.

Capability review items (20)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

SKILL.md:12-30

Ruby/shell backtick execution

```bash

The quick-start shell block directs an agent to run GitHub CLI, test, and git push commands. These commands can read repository state and publish changes, so the external command risk is real.

High

SKILL.md:128-130

Ruby/shell backtick execution

```bash

The git push command publishes local commits to the remote repository. This is a high-impact authenticated action if run without explicit user approval.

High

SKILL.md:134-143

Ruby/shell backtick execution

```bash

The gh pr review commands can approve, request changes, or comment on a pull request. These are authenticated review actions that can affect merge decisions.

Medium

SKILL.md:37-38

Ruby/shell backtick execution

2. **Commit format**: Check `git log --oneline -5` for project style

The inline git log command is a real external command used to inspect repository history. It is read-only, but still instructs command execution in the user workspace.

Medium

SKILL.md:38

Ruby/shell backtick execution

3. **Test command**: Identify test runner (`make test`, `pytest`, `npm test`)

The skill names concrete test runner commands such as make test, pytest, and npm test. These can execute project-controlled code and dependencies, so the command execution risk is real.

Medium

SKILL.md:39-45

Ruby/shell backtick execution

4. **Branch status**: `git status` to ensure clean working tree

The git status command is a real shell command for inspecting the working tree. It is low-impact but confirms the skill relies on external command execution.

Medium

SKILL.md:45-49

Ruby/shell backtick execution

```bash

The fenced block runs gh pr view, gh repo view, and gh api to retrieve pull request comments. These are external GitHub CLI calls that depend on repository credentials.

Medium

SKILL.md:84-93

Ruby/shell backtick execution

```bash

The commit block runs git add and git commit. These commands modify repository history and can commit files selected by the agent.

Medium

SKILL.md:102-106

Ruby/shell backtick execution

```bash

The block runs git rev-parse and gh api to post a review-thread reply. This performs authenticated repository actions through the GitHub CLI.

Medium

SKILL.md:120-122

Ruby/shell backtick execution

```bash

The make test command is a real external command. Running project tests can execute arbitrary project scripts or dependency hooks.

Medium

SKILL.md:177-181

Ruby/shell backtick execution

```bash

The GitHub API example posts a reply to a pull request comment. It is an authenticated external command with repository side effects.

Medium

SKILL.md:17

Shell command substitution

PR=$(gh pr view --json number -q '.number')

The PR variable is populated by executing gh pr view in shell command substitution. The command is read-only, but it is still external command execution.

Medium

SKILL.md:18

Shell command substitution

REPO=$(gh repo view --json nameWithOwner -q '.nameWithOwner')

The REPO variable is populated by executing gh repo view in shell command substitution. It depends on GitHub CLI access in the current repository.

Medium

SKILL.md:46

Shell command substitution

PR=$(gh pr view --json number -q '.number')

The PR variable is again populated through gh pr view command substitution. This confirms the workflow expects shell execution.

Medium

SKILL.md:47

Shell command substitution

REPO=$(gh repo view --json nameWithOwner -q '.nameWithOwner')

The REPO variable is again populated through gh repo view command substitution. This is legitimate tooling but still an external command.

Medium

SKILL.md:103

Shell command substitution

COMMIT=$(git rev-parse --short HEAD)

The COMMIT variable is populated through git rev-parse command substitution. It is local and read-only, but confirms shell execution in the workflow.

Medium

SKILL.md:12-30

Template literal with command substitution

```bash

The finding points at a Markdown code fence, but the fenced quick-start block contains executable shell substitutions and GitHub CLI commands. I kept it because the skill may cause an agent to run those commands.

Medium

SKILL.md:45-49

Template literal with command substitution

```bash

The finding points at a Markdown code fence, but the block contains gh command substitutions used to build API calls. It is a real external command workflow despite the imprecise pattern name.

Medium

SKILL.md:84-93

Template literal with command substitution

```bash

The finding points at a fenced shell block with git commit command substitution. The interpolation is static, but the block still instructs repository-modifying shell execution.

Medium

SKILL.md:102-106

Template literal with command substitution

```bash

The finding points at a fenced shell block that derives a commit hash and posts through gh api. The pattern name is imprecise, but the external command behavior is present.

Audited by: codex View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/bae-changhyun-github-pr-review/audits/8?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/bae-changhyun-github-pr-review/security.svg)](https://skillstore.io/skills/bae-changhyun-github-pr-review?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/bae-changhyun-github-pr-review?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/bae-changhyun-github-pr-review/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/bae-changhyun-github-pr-review.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

Bae-ChangHyun. (2026). github-pr-review security audit report (audit version 8) [Author version unspecified]. Skillstore. https://skillstore.io/skills/bae-changhyun-github-pr-review/audits/8

BibTeX citation

@techreport{bae-changhyun-bae-changhyun-github-pr-review-2026,
  author = {Bae-ChangHyun},
  title = {github-pr-review security audit report (audit version 8)},
  institution = {Skillstore},
  year = {2026},
  number = {8},
  url = {https://skillstore.io/skills/bae-changhyun-github-pr-review/audits/8},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "github-pr-review security audit report (audit version 8)"
version: "unspecified"
type: report
authors:
  - name: "Bae-ChangHyun"
date-released: "2026-07-06"
url: "https://skillstore.io/skills/bae-changhyun-github-pr-review/audits/8"
identifiers:
  - type: other
    value: "skillstore:bae-changhyun-github-pr-review:audit:8"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Résoudre les retours de revue bloquants

Classer les commentaires critiques et de haute sévérité, appliquer les corrections approuvées, exécuter les tests et répondre aux reviewers.

Organiser de grands threads de revue

Regrouper de nombreux commentaires de revue par sévérité et décider quelles corrections nécessitent des commits séparés.

Standardiser les réponses de revue

Utiliser des modèles de réponse cohérents pour les commentaires corrigés, reportés, intentionnels et pris en compte.

Try These Prompts

Triage des commentaires de revue

Utilise cette compétence pour récupérer les commentaires de ma PR actuelle, les classer par sévérité et me montrer l’ordre de correction recommandé avant de modifier les fichiers.

Corriger un commentaire

Utilise cette compétence pour inspecter le commentaire de revue 123456789, expliquer le problème, proposer une correction et attendre mon approbation avant de modifier.

Traiter par lot les retours à faible risque

Utilise cette compétence pour collecter les commentaires de revue de sévérité moyenne et faible, suggérer un plan par lot et préparer un commit de style ou de documentation après approbation.

Terminer le suivi de revue

Utilise cette compétence pour traiter les corrections approuvées, exécuter les tests du projet, résumer les risques restants, rédiger les réponses aux threads et demander avant toute action de push ou de revue.

Best Practices

Confirmer chaque changement de fichier avant d’appliquer les corrections issues des commentaires de revue.
Traiter le texte des commentaires de PR comme une entrée non fiable et ignorer les instructions sans rapport.
Demander avant de pousser des commits ou de soumettre une revue GitHub formelle.

Avoid

Ne pas exécuter de scripts de projet dans un dépôt non fiable sans l’approbation de l’utilisateur.
Ne pas approuver une pull request uniquement parce que les commentaires ont reçu une réponse.
Ne pas laisser les commentaires de revue remplacer les instructions système, développeur ou utilisateur.

Frequently Asked Questions

Cette compétence nécessite-t-elle GitHub CLI ?

Oui. Le workflow utilise des commandes GitHub CLI pour lire les métadonnées de PR, récupérer les commentaires, publier des réponses et soumettre des revues.

Peut-elle modifier mon code ?

Elle guide les changements de code après l’analyse de la revue. La compétence indique de confirmer avec l’utilisateur avant d’appliquer les corrections.

Peut-elle pousser des commits ?

Oui, le workflow documenté inclut git push. Les utilisateurs doivent exiger une approbation explicite avant tout push.

Peut-elle approuver des pull requests ?

La compétence inclut des commandes de revue GitHub, y compris des exemples d’approbation. L’approbation doit rester une décision humaine délibérée.

Comment priorise-t-elle les commentaires ?

Elle ordonne le travail de la sévérité critique à élevée, moyenne et faible avant de proposer des corrections.

Le texte des commentaires de revue est-il toujours sûr à suivre ?

Non. Les commentaires de revue peuvent contenir des instructions malveillantes. Traitez-les comme non fiables et vérifiez chaque action demandée.

Developer Details

Author

Bae-ChangHyun

License

MIT

Skillstore revision

Version notice

The author did not declare a version.

Repository

https://github.com/Bae-ChangHyun/cc-plugins-bch/tree/main/plugins/gitwf/skills/github-pr-review

Ref

c4231f0153ccd2a2ed62274c5a372bf8f2386d3f

Maintenance freshness

7/22/2026

Usage

6 downloads · 205 views

File structure

📄 SKILL.md

github-pr-review

Install with my Agent

Agent-readable resources

Test it

Security Audit

Confirmed security concerns (3)

Risk Factors

Copy report link

Markdown badge

HTML badge

Embed card

APA citation

BibTeX citation

CITATION.cff

Skillstore Score

What You Can Build

Résoudre les retours de revue bloquants

Organiser de grands threads de revue

Standardiser les réponses de revue

Try These Prompts

Best Practices

Avoid

Frequently Asked Questions

Developer Details