pr-review

Coordinate four parallel GitHub PR review agents with optional automation.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/so-ota-biz/ai-agent-orchestration-settings --skill pr-review-so-ota-biz
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-review
Source: https://github.com/so-ota-biz/ai-agent-orchestration-settings/tree/main/.agent/skills/pr_review
Command: npx skills add https://github.com/so-ota-biz/ai-agent-orchestration-settings --skill pr-review-so-ota-biz

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

GitHub PR reviews can be time-consuming and error-prone when coordinating multiple perspectives; this skill orchestrates a four-agent harness to complete reviews efficiently.

Core Features & Use Cases

  • Parallel review harness: launches code-reviewer, code-quality-reviewer, must-check-reviewer, and review-meta-checker in independent contexts to avoid cross-contamination of viewpoints.
  • Controlled automation: supports --auto to automatically post only when critical issues are detected, otherwise prompts for user confirmation.
  • Comprehensive coverage: ensures review coverage across general quality, language/style, must-check criteria, and meta-quality checks; supports PR URL or number input.

Quick Start

Provide a GitHub PR number or URL to start the four-agent harness 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 pull request reviews using multiple agents?

Automate GitHub pull request reviews by providing a PR number or URL to a four-agent harness, which coordinates parallel reviews across code, quality, must-check criteria, and meta-checks for comprehensive assessment.

What is a multi-agent harness for code review and how does it work?

A multi-agent harness for code review launches four independent agents—code-reviewer, code-quality-reviewer, must-check-reviewer, and review-meta-checker—in isolated contexts to prevent cross-contamination of viewpoints during parallel execution.

How do I start a parallel PR review with a specific PR number?

Start a parallel PR review by passing the GitHub PR number or URL as input to trigger the four-agent harness, which then executes independent review contexts automatically.

Can I automatically post review comments only when critical issues are detected?

Yes, you can automatically post review comments only when critical issues are detected by using the --auto setting, which otherwise prompts for user confirmation before posting.

Does the PR review harness support custom automation effort levels?

Yes, the PR review harness supports custom automation effort levels through the --effort setting, allowing controlled automation during the post-review loop.

What are the limitations of using a single-context code review compared to parallel agents?

Single-context code reviews risk cross-contamination of viewpoints, whereas a parallel four-agent harness isolates general quality, language/style, must-check, and meta-quality checks into independent contexts for comprehensive coverage.