Does this skill modify my code automatically?

No, Team Review is analysis-only. It identifies issues and suggests fixes, but does not modify files. Use /implement-feature or /fix-bug to apply suggested changes.

How do I focus the review on just one specific area?

You can use the --focus flag followed by an area like 'security', 'performance', or 'architecture' to spawn only the reviewers relevant to that domain.

Can I use this skill for an entire repository or just PRs?

It supports both. You can provide a PR number, a specific commit SHA, or use the --all flag to review the entire current codebase.

What is the difference between Full and Lite mode?

Full mode uses the experimental Teammate API to spawn 7 distinct agents, while Lite mode uses Task subagents with combined roles for better cost efficiency and compatibility.

What is the role of the Adversary Reviewer?

The Adversary agent specifically waits for initial findings to challenge them, looking for false positives, untested assumptions, and extreme edge cases at 10x scale.

Team Review

Name: Team Review
Author: mgiovani

bymgiovani

•

安全与测试

Orchestrates a multi-agent team of specialized AI reviewers to perform comprehensive security, architectural, and performance code reviews.

Team Review is a premium Claude Code skill designed for high-stakes pull requests and critical architectural changes. It coordinates up to seven specialized agents—covering Architecture, Security, Performance, Testing, Style, Docs/UX, and a unique Adversary agent—to provide a level of depth impossible for a single-agent review. By automatically discovering project conventions and stress-testing findings with an adversary model, it ensures production-grade code quality while minimizing false positives through evidence-based reporting.

主要功能

012 GitHub stars

02Automated Project Discovery: Detects technology stacks, coding conventions, and quality standards before starting.

03Adversary Reviewer: A dedicated agent that challenges assumptions, finds edge cases, and validates other agents' findings.

04Dual Operation Modes: Supports experimental 'Agent Teams' for maximum depth or 'Lite Mode' for cost-efficient Task subagents.

05Evidence-Based Reporting: Generates structured reports with file/line references, severity ratings, and actionable fix suggestions.

06Multi-Agent Orchestration: Spawns 7 specialized reviewers for deep, multi-dimensional analysis.

使用场景

01Auditing security-sensitive PRs involving authentication, payments, or PII handling.

02Reviewing major architectural refactors or new system designs for scalability and reliability.

03Comprehensive code quality assessments before major production releases or version tags.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mgiovani/cc-arsenal team-review

For use in Claude.ai and ChatGPT

Download Skill