code-review

Coordinate multi-agent pull request reviews with parallel adversarial and operational analysis.

9|1|Updated Mar 26, 2026

Search Tags:#code review #subagents #multi-agent #pull requests #triage #copilot #production risk

Ready to run

npx skills add https://github.com/conorbronsdon/claude-code-skills --skill code-review-conorbronsdon

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/conorbronsdon/claude-code-skills/tree/main/code-review
Command: npx skills add https://github.com/conorbronsdon/claude-code-skills --skill code-review-conorbronsdon

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you perform deeper pull request reviews than a single pass by coordinating Copilot with parallel subagents to catch architectural misses, security flaws, and operational risks before merge.

Core Features & Use Cases

Scope-based review: Classifies pull requests as trivial, standard, or high-stakes and adjusts review depth accordingly.
Parallel subagent analysis: Runs adversarial, operational, and reference-comparison perspectives at the same time to surface different classes of issues.
Triage and iteration control: Separates real findings from stale re-flags and caps review rounds to keep signal high.
Use case: Ideal for payment, auth, deployment, SDK integration, or any PR where production risk is higher than a normal code review can safely cover.

Quick Start

Use the code-review skill to assess the PR, run the appropriate subagents in parallel, request Copilot review, and triage findings with the two-round cap.

Frequently Asked Questions about code-review

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

FAQPage Schema

How do I run a multi-agent pull request review to find hidden production risks?▼

A multi-agent pull request review coordinates Copilot with parallel subagents to expose architectural misses, security flaws, and operational risks. It classifies PRs by scope and applies adversarial operational analysis to surface critical defects that single-pass reviews miss.

What's the best way to review high-stakes code changes before merge?▼

Reviewing high-stakes code changes requires scope-based escalation and parallel subagent analysis to catch production risks. This approach applies adversarial, operational, and reference-comparison perspectives simultaneously to ensure comprehensive coverage for security-sensitive PRs.

Can I use this code review process for payment and auth pull requests?▼

Yes, this code review process is ideal for payment, auth, deployment, and SDK integration pull requests. It adjusts review depth based on scope classification, ensuring high-stakes and third-party integration PRs receive the necessary adversarial and operational analysis.

How does stale-finding triage work during parallel subagent analysis?▼

Stale-finding triage separates real findings from outdated re-flags during parallel subagent analysis. It works alongside a strict two-round iteration cap to keep the review signal high and prevent redundant checks from overwhelming the final defect report.

Why does a single-pass code review miss architectural defects?▼

A single-pass code review misses architectural defects because it lacks parallel adversarial and operational perspectives. Coordinating multiple subagents simultaneously exposes different classes of issues like security flaws and operational risks that a sequential single-pass review cannot safely cover.