review-pr

Fetch GitHub pull requests, create worktrees, and run parallel review agents.

Updated Sep 9, 2024
One-click install
npx skills add https://github.com/axel-kaliff/dotfiles --skill review-pr-axel-kaliff
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review-pr
Source: https://github.com/axel-kaliff/dotfiles/tree/main/claude/.claude/skills/review-pr
Command: npx skills add https://github.com/axel-kaliff/dotfiles --skill review-pr-axel-kaliff

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Pull request reviews are time-consuming and can miss behavioral regressions, historical context, and repository-specific rules; this Skill automates fetching the PR, creating an isolated worktree, running a structured set of review agents, and surfacing concise, high-confidence findings so reviewers can act faster.

Core Features & Use Cases

  • Worktree isolation: fetches the PR branch and creates or reuses a local worktree so reviewers can inspect the code locally without disrupting their current workspace.
  • Multi-agent analysis: runs parallel agents to check CLAUDE.md compliance, shallow bug scans, git-history context, prior PR comments, code-comments policy, and deep per-file behavioral comparisons.
  • Scoring and posting: consolidates findings, scores them with a dedicated scoring sub-skill, and drafts a concise Swedish PR comment for high-confidence issues that the user may post with approval.
  • Use Case: a reviewer needs an end-to-end, behavior-aware review of a feature branch that may subtly change control flow or exception handling across multiple files.

Quick Start

Run the review-pr skill on PR 1889 to perform a full multi-agent review and produce a draft Swedish comment for posting.

Frequently Asked Questions about review-pr

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate a GitHub pull request review with behavioral analysis?

Automated GitHub pull request reviews use multi-agent analysis to fetch the PR, create an isolated worktree, and run parallel agents that check for behavioral regressions, CLAUDE.md compliance, and historical context before drafting a summarized comment.

What is a multi-agent PR review and how does it work?

A multi-agent PR review runs parallel agents to perform shallow bug scans, git-history context analysis, prior PR comment checks, and deep per-file behavioral comparisons, then consolidates and scores findings for high-confidence issues.

Do I need git and gh CLI access to run automated code reviews?

Yes, automated code reviews require git and gh CLI access along with repository read/write permissions to create local worktrees, fetch PR branches, and draft comments without disrupting your current workspace.

Can I review a feature branch without disrupting my current git workspace?

Yes, worktree isolation fetches the PR branch and creates or reuses a local worktree so you can inspect code and run behavioral checks locally without disrupting your current workspace.

How to check pull request compliance with repository-specific rules?

Pull request compliance checking uses parallel review agents to verify CLAUDE.md rules, code-comments policy, and prior PR context, then scores findings with a dedicated sub-skill for deduplication and ranking.

What are the limitations of automated multi-agent code reviews?

Automated multi-agent code reviews focus on high-confidence findings for draft comments and require user approval before posting, meaning low-confidence issues may be omitted and the Swedish-language draft comment may need manual translation.