github-pr-review

tolgaio 的头像作者 tolgaio

提取并全面分析GitHub拉取请求

📖 中文导读:提取并全面分析GitHub拉取请求 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: github-pr-review description: Extract and review GitHub Pull Requests with comprehensive analysis

GitHub PR Review Skill

Purpose

This skill extracts comprehensive GitHub Pull Request data and provides a structured methodology for code review. It fetches PR metadata, commits, comments, reviews, file changes, and diffs via the GitHub API, then guides analysis across code quality, security, testing, and best practices.

When to Use

Use this skill when you need to:

  • Review a GitHub Pull Request
  • Extract detailed PR information for analysis
  • Generate a comprehensive PR summary
  • Provide structured code review feedback

Prerequisites

Required Tools

  • curl - for GitHub API requests
  • jq - for JSON parsing
  • git - for repository operations

Environment

  • GITHUB_TOKEN - GitHub personal access token (optional for public repos, required for private)

Usage

Use the pr-reviewer agent for context isolation:

/review-pr https://github.com/owner/repo/pull/123

Direct Skill Invocation

Follow the instructions in instructions.md:

  1. Run extraction:
$HOME/.claude/skills/github-pr-review/scripts/extract.sh \
  --owner <owner> --repo <repo> --pr <number>
  1. Read the generated extract.md from the output directory

  2. Apply the review methodology from instructions.md

  3. Save the review to review.md in the same directory

What Gets Extracted

Data Source
PR metadata Title, author, URL, description
Commit history All commits with SHA, title, description, author
Comments General PR comments with timestamps
Review comments Line-specific comments with file:line context
Reviews Approval states (APPROVED, CHANGES_REQUESTED, etc.)
File changes Diffs and full post-change content
Documentation links Auto-detected based on file types

Output

The skill creates a directory structure for each PR review:

$BRAIN_HOME/github-pr-reviews/{owner}/{repo}/pr-{number}/
├── extract.md    # Raw PR data from GitHub API
└── review.md     # Structured analysis with summary, findings, and verdict

This organization:

  • Keeps all PR reviews in a central location
  • Preserves review history for future reference
  • Groups reviews by repository for easy navigation

Integration

  • Agent: agents/pr-reviewer.md - runs skill in isolated context
  • Command: commands/review-pr.md - invokes the agent