pr-review

Classify PR reviews and create audit trails from GitHub diffs via gh CLI.

3|Updated Mar 17, 2026
One-click install
npx skills add https://github.com/marcoguillermaz/claude-dev-kit --skill pr-review-marcoguillermaz
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-review
Source: https://github.com/marcoguillermaz/claude-dev-kit/tree/main/packages/cli/templates/tier-m/.claude/skills/pr-review
Command: npx skills add https://github.com/marcoguillermaz/claude-dev-kit --skill pr-review-marcoguillermaz

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates PR review classification and audit trail creation by analyzing PR diffs and metadata.

Core Features & Use Cases

  • Fetches PR metadata and diffs from GitHub using the gh CLI.
  • Applies universal severity defaults and stack-specific patterns from PATTERNS.md, with optional team overrides.
  • Spawns a review subagent to generate a structured markdown review and posts it as a PR comment for an auditable trail.

Quick Start

Run /pr-review <PR_NUMBER> in the local CLI to begin an autonomous PR review.

Frequently Asked Questions about pr-review

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

FAQPage Schema
How do I automate GitHub PR review and create an audit trail?

Automating GitHub PR review involves fetching PR diffs and metadata via the gh CLI, applying severity criteria, and posting a structured markdown review comment. This process operates read-only to avoid code changes while generating an auditable review trail.

What is an autonomous PR review orchestrator and how does it work?

An autonomous PR review orchestrator analyzes PR diffs and metadata using universal severity defaults and stack-specific patterns. It reads project conventions from CLAUDE.md and PATTERNS.md, then spawns a subagent to generate structured markdown reviews posted as PR comments.

Can I customize PR review severity criteria with team overrides?

Yes, PR review severity criteria can be customized using team overrides. The review process applies universal severity defaults and stack-specific patterns from PATTERNS.md, while optional team overrides allow project-specific customization for the audit trail.

Does the PR review process require write access to the repository?

No, the PR review process operates in a read-only orchestration to avoid code changes. It uses the gh CLI to fetch PR metadata and diffs, analyzes them against severity patterns, and posts review comments without modifying the repository codebase.

How do I run an autonomous PR review locally using the gh CLI?

To run an autonomous PR review locally, execute the review command with a PR number in your local CLI. The orchestrator fetches the PR diff and metadata via gh CLI, analyzes it against configured patterns, and posts a structured markdown review comment.

What's the best way to classify PR diff severity using stack-specific patterns?

The best way to classify PR diff severity is applying universal severity defaults alongside stack-specific patterns from PATTERNS.md. The orchestrator reads project conventions from CLAUDE.md and uses optional team overrides to ensure accurate, project-customized severity classification.