code-review

Analyze codebases for T-stage reviews using LLM-as-Judge and output structured JSON.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/isaenter/UES-PRO --skill code-review-isaenter
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/isaenter/UES-PRO/tree/main/frontend-vben/.claude/skills/code-review
Command: npx skills add https://github.com/isaenter/UES-PRO --skill code-review-isaenter

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

T 阶段代码审查 — 含 LLM-as-Judge 模式,帮助团队自动化、统一化地评估代码质量、合规性与安全性,减少人工偏差。

Core Features & Use Cases

  • LLM-assisted evaluation: 以 LLM-as-Judge 的模式对代码进行可解释的评审与决策。
  • Checklist-driven review: 检查 Spec 合规、TDD 覆盖、边界与错误路径处理、数据安全和代码可读性等要点。
  • Delivery-ready outputs: 产出结构化的 JSON 报告,包含 level、issues 与 summary,便于集成到 CI/CD 流程。
  • Use Case: 在新特性提交、变更审查、以及安全合规审查中提供一致的评审结果。

Quick Start

Run the skill on a codebase to generate a JSON report in the LLM-as-Judge format.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I automate code review for spec compliance and testing coverage?

Automate code review by running an LLM-as-Judge skill on your codebase to evaluate spec compliance, testing coverage, boundary handling, and security. It outputs a structured JSON report containing severity level, identified issues, and a summary for consistent quality assurance.

What is the LLM-as-Judge pattern for static analysis and code review?

The LLM-as-Judge pattern applies large language models to evaluate code quality, security, and spec compliance against a predefined rubric. This approach provides explainable review decisions and structured outputs, reducing human bias in T-stage code assessments.

Can I integrate automated code review results into CI/CD pipelines?

Yes, you can integrate automated code review results into CI/CD pipelines. The skill generates delivery-ready structured JSON with predefined fields like level, issues, and summary, making it straightforward to parse and act upon in automated workflows.

How does LLM-as-Judge compare to traditional static analysis tools for security and quality assurance?

Unlike traditional static analysis tools that rely on fixed rules, LLM-as-Judge evaluates codebases holistically for spec compliance, boundary handling, and readability. It provides explainable assessments validated against a predefined rubric, reducing manual bias across reviews.

Do I need to configure a specific checklist for boundary handling and data security reviews?

No specific configuration is needed. The checklist-driven review automatically assesses spec compliance, TDD coverage, boundary and error path handling, data security, and code readability across typical projects without requiring manual checklist setup.