skills.rest
unorouter
unorouterOFFICIAL
Explore all skills by this author →
WebsiteGitHub

code-review-expert

Coordinate multi-agent code reviews using isolated git worktrees and generate Markdown reports.

Updated Feb 12, 2026
Search Tags:#architecture#security#multi-agent#code-review#context7#git-worktree
One-click install
npx skills add https://github.com/unorouter/sub2api --skill code-review-expert-unorouter
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: code-review-expert
Source: https://github.com/unorouter/sub2api/tree/main/skills/code-review-expert
Command: npx skills add https://github.com/unorouter/sub2api --skill code-review-expert-unorouter

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

通用代码审核专家通过多代理并行审核和 Context7 MCP 三重验证来对代码库进行全面质量、安全性、可靠性评估，同时确保不修改原始代码。

Core Features & Use Cases

  • 多代理并行审核：自动创建 5 个 git worktree，分发到独立子 Agent 并行审核，提升效率与隔离性。
  • Context7 防幻觉验证：Phase 0/2/3 三层验证，拉取官方文档以消除模型幻觉。
  • 自动报告与清理：汇总为结构化 Markdown 审核报告 code-review-report.md，最终自动清理所有工作区。
  • 适用于对 PR、分支、目录或文件进行全面质量检查，覆盖安全合规、架构设计、性能资源、可靠性与数据完整性、可观测性等五大维度。

Quick Start

Provide the repository to audit and trigger the code-review-expert workflow to begin parallel, context-verified code reviews.

Frequently Asked Questions about code-review-expert

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

FAQPage Schema
How do I run parallel code reviews on a PR across multiple dimensions?▼

Parallel code reviews use 5 isolated git worktrees dispatched to independent sub-agents to audit quality, security, and reliability dimensions simultaneously without modifying the original codebase.

What is Context7 MCP verification in a multi-agent code review?▼

Context7 MCP verification is a three-phase anti-hallucination mechanism that pulls official documentation during the code review process to eliminate model hallucinations and ensure accurate quality assessments.

Can I audit specific directories or files instead of an entire branch?▼

Yes, the code review workflow can be applied to PRs, branches, directories, or individual files across any tech stack, performing comprehensive audits and generating structured Markdown reports for the targeted scope.

Does the parallel code review process modify my original code?▼

No, the code review process does not modify your original code because all analysis occurs within isolated git worktrees that are automatically cleaned up after generating the structured Markdown report.

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

Limitations include enforced Context7 MCP verification requirements and restricted file read limits to control scope, meaning extremely large repositories may require targeted directory or file-level audits rather than full branch reviews.

Related Skills

Explore More Software Engineering Skills→
pradeepmoulipradeepmouli
2

speckit-worktree-create

Create isolated git worktrees for feature branches under .worktrees/.

Community
Advanced
mt-krainskimt-krainski
1

wrap

Automate branch completion with code review, linting, testing, commits, and Pull Request creation.

Community
Advanced
LLM-Dev-OpsLLM-Dev-Ops
1

github-code-review

Automate GitHub pull request reviews with multi-agent swarm analysis.

Official
Advanced
YuhangNieYuhangNie
3

code-review-expert

Perform multi-agent code reviews across security, architecture, performance, reliability, and quality.

Community
Advanced