review-pr

スター3

フォーク0

更新日2026年4月3日 16:06

Review a pull request using parallel agents for thorough coverage

インストール

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

ソース

przadka

przadka/configure-cc

ダウンロード

関連職種SOC

SOC 職業分類に基づく

SKILL.md

readonly

name	review-pr
description	Review a pull request using parallel agents for thorough coverage
disable-model-invocation	true
argument-hint	[PR number or URL]

Review pull request $ARGUMENTS using parallel review agents.

Fetch the PR diff: gh pr diff $ARGUMENTS
Spawn 3 parallel review agents:
- Correctness: logic bugs, edge cases, error handling gaps
- Security: injection risks, auth issues, data exposure
- Style: naming, consistency with existing patterns, unnecessary complexity
Collect findings into a single triaged list:
- P1 (must fix): bugs, security issues, data loss risks
- P2 (should fix): logic gaps, missing validation, poor error messages
- P3 (nitpick): style, naming, minor improvements
Present the consolidated list and ask which issues to address
When posting to GitHub, use CHANGES_REQUESTED (not COMMENT) if P1/P2 issues exist
Always create secret gists (never public) for shared artifacts

このリポジトリの他の Skills

同じリポジトリ

Audit a project's CLAUDE.md / AGENTS.md for structure, bloat, stale references, and best practices

Read-only analysis of global Claude Code configuration — flags issues and suggests improvements

Set up global Claude Code configuration — personal CLAUDE.md, skills, hooks, and settings

Scan the current project and generate tailored project-level Claude Code configuration

Create a git commit with proper context and conventional commit message

name	review-pr
description	Review a pull request using parallel agents for thorough coverage
disable-model-invocation	true
argument-hint	[PR number or URL]

Review pull request $ARGUMENTS using parallel review agents.

Fetch the PR diff: gh pr diff $ARGUMENTS
Spawn 3 parallel review agents:
- Correctness: logic bugs, edge cases, error handling gaps
- Security: injection risks, auth issues, data exposure
- Style: naming, consistency with existing patterns, unnecessary complexity
Collect findings into a single triaged list:
- P1 (must fix): bugs, security issues, data loss risks
- P2 (should fix): logic gaps, missing validation, poor error messages
- P3 (nitpick): style, naming, minor improvements
Present the consolidated list and ask which issues to address
When posting to GitHub, use CHANGES_REQUESTED (not COMMENT) if P1/P2 issues exist
Always create secret gists (never public) for shared artifacts