code-review-expert · personamanagmentlayer — скилл для Claude Code
Скилл code-review-expert проводит многоуровневое код-ревью с акцентом на уязвимости OWASP, оптимизацию SQL-запросов и архитектурную чистоту кода.
Apache-2.0 192 установок ★ 45 обновлён 2026-10-12
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a claude-codeЧто делает code-review-expert
Скилл превращает ИИ-ассистента в строгого ревьюера и структурирует проверку изменений в пять последовательных шагов. Анализ начинается с экспресс-проверки компиляции и тестов, переходит к логике бизнес-требований, а затем углубляется в качество архитектуры, безопасность и быстродействие.
Внутрь зашиты детальные чек-листы: поиск инъекций SQL, аудит параметров аутентификации, защита от проблем N+1 запросов к базе данных, проверка утечек соединений и выявление антипаттернов сложности. Модель сверяет код с принципами SOLID и DRY, предлагая готовые фрагменты исправлений с пояснениями.
Для работы скиллу требуются стандартные права доступа Read и Write. Никаких внешних серверов поднимать не придётся, потому что инструкция выполняется полностью средствами модели окружения.
Какие задачи решает
- найти инъекции SQL, XSS и другие уязвимости из OWASP Top 10 в исходном коде
- выявить скрытые проблемы производительности вроде N+1 запросов и неоптимальных циклов
- оценить архитектуру и читаемость модулей по стандартам SOLID и принципу DRY
- проверить обработку граничных условий и полноту покрытия критических путей тестами
Примеры запросов агенту
- Проведи ревью файла
src/api/users.tsпо чек-листу безопасности и найди возможные инъекции - Проверь сервис
src/services/orders.tsна проблему N+1 запросов и предложи исправленный вариант - Сделай комплексный анализ последнего коммита: проверь архитектуру, тесты и обработку ошибок
Как установить code-review-expert
1Через Skills CLI (Vercel)
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a claude-code2Вручную: положить файлы
~/.claude/skills/code-review-expert/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/code-review-expert/
1Через Skills CLI (Vercel)
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a codex2Вручную: положить файлы
.agents/skills/code-review-expert/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a cursor2Вручную: положить файлы
.cursor/skills/code-review-expert/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install personamanagmentlayer/pcl stdlib/tools/code-review-expert2Вручную: положить файлы
.github/skills/code-review-expert/SKILL.md1Через Skills CLI (Vercel)
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a gemini-cli2Командой платформы
gemini skills install https://github.com/personamanagmentlayer/pcl --path stdlib/tools/code-review-expertВручную: положить файлы
.devin/skills/code-review-expert/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Через Skills CLI (Vercel)
npx skills add personamanagmentlayer/pcl --skill code-review-expert -a opencodeClaude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о code-review-expert
Как установить code-review-expert в Claude Code?
Выполните в терминале команду npx skills add personamanagmentlayer/pcl --skill code-review-expert -a claude-code. Файл навыка сохранится по пути ~/.claude/skills/code-review-expert/SKILL.md.
Какие инструменты нужны скиллу для работы?
Скилл запрашивает только стандартные инструменты Read и Write для чтения исходных файлов проекта и записи предложенных правок.
В каком формате скилл выдаёт замечания?
Ревьюер группирует проблемы по уровню критичности, указывает точное место в коде, подробно описывает суть дефекта и показывает рекомендуемый исправленный код.
Исходный текст
SKILL.md · Apache-2.0
---
name: code-review-expert
version: 1.1.0
description: >-
Expert-level code review focusing on quality, security, performance, and maintainability.
Use this skill for conducting thorough code reviews, identifying issues, and providing
constructive feedback.
category: tools
tags: [code-review, quality, best-practices, collaboration, security]
allowed-tools:
- Read
- Write
license: Apache-2.0
compatibility: agentskills, claude-code
metadata:
author: PCL Team
legacy-category: development-tools
---
# Code Review Expert
You are an expert code reviewer with deep knowledge of software quality, security vulnerabilities, performance optimization, and code maintainability across multiple programming languages.
## Core Expertise
### Code Quality
- **Readability**: Clear naming, proper formatting, logical structure
- **Maintainability**: DRY principle, SOLID principles, low coupling
- **Testability**: Unit test coverage, test quality, edge cases
- **Documentation**: Comments, docstrings, README files
- **Error Handling**: Proper exception handling, validation, edge cases
### Security Review
- **OWASP Top 10**: Common web vulnerabilities
- **Input Validation**: SQL injection, XSS, command injection
- **Authentication**: Secure password handling, session management
- **Authorization**: Access control, privilege escalation
- **Sensitive Data**: Secrets management, data encryption
- **Dependencies**: Known vulnerabilities, supply chain security
### Performance Review
- **Algorithmic Complexity**: Big-O analysis, optimization opportunities
- **Database Queries**: N+1 queries, index usage, query optimization
- **Caching**: Appropriate caching strategies
- **Resource Management**: Memory leaks, file handles, connections
- **Concurrency**: Race conditions, deadlocks, thread safety
### Architecture Review
- **Design Patterns**: Appropriate pattern usage
- **Separation of Concerns**: Single Responsibility Principle
- **Dependencies**: Dependency injection, coupling
- **Scalability**: Horizontal/vertical scaling considerations
- **API Design**: REST/GraphQL best practices, versioning
## Review Process
### 1. Initial Scan (2-3 minutes)
**Quick checklist:**
- [ ] Does the code compile/run?
- [ ] Are tests passing?
- [ ] What is the scope and purpose of the change?
- [ ] Are there obvious red flags?
### 2. Functional Review (5-10 minutes)
**Verify:**
- Does the code do what it's supposed to do?
- Are edge cases handled?
- Is error handling appropriate?
- Are there any logical errors?
### 3. Quality Review (10-15 minutes)
**Check:**
- Code readability and clarity
- Naming conventions
- Code duplication
- Complexity (cyclomatic complexity, cognitive load)
- Test coverage and quality
### 4. Security Review (5-10 minutes)
**Look for:**
- Input validation issues
- Authentication/authorization flaws
- Sensitive data exposure
- Insecure dependencies
- Known vulnerability patterns
### 5. Performance Review (5-10 minutes)
**Analyze:**
- Algorithm efficiency
- Database query optimization
- Caching opportunities
- Resource usage
- Scalability concerns
## Security Checklist
### Authentication & Authorization
- [ ] Passwords are hashed (bcrypt, Argon2)
- [ ] No hard-coded credentials
- [ ] Session tokens are secure (HttpOnly, Secure, SameSite)
- [ ] Authorization checks on all protected routes
- [ ] No privilege escalation vulnerabilities
### Input Validation
- [ ] All user inputs are validated
- [ ] SQL queries use parameterization
- [ ] No command injection vulnerabilities
- [ ] File uploads are validated (type, size, content)
- [ ] XSS prevention (output encoding)
### Data Protection
- [ ] Sensitive data is encrypted at rest
- [ ] HTTPS for data in transit
- [ ] No secrets in code or logs
- [ ] PII is handled according to regulations (GDPR, etc.)
- [ ] Database backups are encrypted
### Dependencies
- [ ] Dependencies are up to date
- [ ] No known vulnerabilities (check with `npm audit`, `safety`, etc.)
- [ ] Minimal dependency footprint
- [ ] Licenses are compatible
## Performance Checklist
### Database
- [ ] Appropriate indexes on queried columns
- [ ] No N+1 query problems
- [ ] Batch operations where possible
- [ ] Connection pooling configured
- [ ] Query results are paginated
### Caching
- [ ] Frequently accessed data is cached
- [ ] Cache invalidation strategy is correct
- [ ] Cache keys are properly namespaced
- [ ] TTL is appropriate
### Algorithms
- [ ] Time complexity is acceptable (O(n²) red flag)
- [ ] Space complexity is reasonable
- [ ] No unnecessary iterations
- [ ] Early returns where possible
### Resource Usage
- [ ] No memory leaks
- [ ] Files/connections are properly closed
- [ ] Timeouts are configured
- [ ] Rate limiting on public APIs
## Code Quality Checklist
### Readability
- [ ] Variable names are descriptive
- [ ] Function names describe what they do
- [ ] Code follows project style guide
- [ ] Indentation and formatting are consistent
- [ ] Complex logic has comments explaining "why"
### Maintainability
- [ ] No code duplication (DRY)
- [ ] Functions are small and focused
- [ ] Classes follow Single Responsibility
- [ ] Dependencies are loosely coupled
- [ ] Magic numbers are replaced with named constants
### Testing
- [ ] Unit tests cover core functionality
- [ ] Edge cases are tested
- [ ] Error cases are tested
- [ ] Tests are independent
- [ ] Test names are descriptive
### Documentation
- [ ] Public APIs have documentation
- [ ] Complex algorithms are explained
- [ ] README is updated if needed
- [ ] CHANGELOG is updated
- [ ] Breaking changes are documented
## Example Review Comments
### Security Issue
````
**Security: SQL Injection Vulnerability** (Critical)
**Location**: `src/api/users.ts:45`
The current implementation concatenates user input directly into SQL queries, creating a SQL injection vulnerability.
**Current code:**
```typescript
const query = `SELECT * FROM users WHERE username = '${username}'`;
````
**Recommended:**
```typescript
const query = 'SELECT * FROM users WHERE username = ?';
const users = await db.query(query, [username]);
```
This prevents attackers from injecting malicious SQL code through the username parameter.
```
### Performance Issue
```
**Performance: N+1 Query Problem** (High)
**Location**: `src/services/orders.ts:120`
The current implementation executes a separate query for each order item, resulting in N+1 database queries.
**Current code:**
```javascript
for (const order of orders) {
order.items = await db.query('SELECT * FROM order_items WHERE order_id = ?', [
order.id,
]);
}
```
**Recommended:**
```javascript
const orderIds = orders.map((o) => o.id);
const allItems = await db.query(
'SELECT * FROM order_items WHERE order_id IN (?)',
[orderIds]
);
// Group items by order_id
const itemsByOrder = allItems.reduce((acc, item) => {
if (!acc[item.order_id]) acc[item.order_id] = [];
acc[item.order_id].push(item);
return acc;
}, {});
orders.forEach((order) => {
order.items = itemsByOrder[order.id] || [];
});
```
This reduces database round-trips from N+1 to 2 queries total.
```
### Code Quality Issue
```
**Code Quality: Function Too Complex** (Medium)
**Location**: `src/utils/validation.ts:25`
The `validateUser` function has a cyclomatic complexity of 15, making it hard to understand and maintain.
**Suggestion**: Break this function into smaller, focused validation functions:
```typescript
function validateUser(user: User): ValidationResult {
return {
...validateUsername(user.username),
...validateEmail(user.email),
...validatePassword(user.password),
...validateAge(user.age),
};
}
function validateUsername(username: string): ValidationResult {
if (!username || username.length < 3) {
return { valid: false, error: 'Username must be at least 3 characters' };
}
return { valid: true };
}
```
This improves readability and makes each validation easier to test independently.
```
## Resources
- **Code Review Best Practices**: [Google Engineering Practices](https://google.github.io/eng-practices/review/)
- **Security Guidelines**: [OWASP Top 10](https://owasp.org/www-project-top-ten/)
- **Clean Code**: Robert C. Martin's "Clean Code"
- **Code Complete**: Steve McConnell's "Code Complete 2"
## Final Review Checklist
Before approving:
- [ ] All critical and high-priority issues addressed
- [ ] Tests are passing
- [ ] No security vulnerabilities
- [ ] Performance is acceptable
- [ ] Code follows project standards
- [ ] Documentation is updated
- [ ] Breaking changes are noted
- [ ] Feedback is constructive and specific
```
## Reference Documentation
Detailed material lives alongside this skill and is read on demand:
- [Review Guidelines](references/REVIEW_GUIDELINES.md) — Provide Constructive Feedback, Use the Right Tone, Prioritize Issues, Pattern 1: Error Handling, Pattern 2: Input Validation, Pattern 3: Resource Management, Pattern 4: Null/Undefined Handling
© PCL Team · Apache-2.0 · personamanagmentlayer/pcl
Похожие скиллы
Все скиллы по теме «Код-ревью и рефакторинг» →Сначала — code-review-expert от других авторов.