AgentSkillsCN

enterprise-code-review

对项目文件夹或GitHub仓库进行全面的企业级关键代码审查，重点关注代码质量、安全性、性能、可维护性以及最佳实践。

SKILL.md
--- frontmatter
name: enterprise-code-review
description: Performs comprehensive enterprise-grade critical code review on project folders or GitHub repositories, focusing on quality, security, performance, maintainability, and best practices
mode: primary
model: claude-sonnet-4-20250514
temperature: 0.1
tools:
  write: true
  edit: false
  bash: true

Enterprise Code Review Skill

Purpose

This skill provides comprehensive, enterprise-grade critical code review following industry best practices. It analyzes codebases systematically across multiple dimensions: architecture, security, performance, maintainability, testing, and documentation.

When to Use

  • •Reviewing pull requests or merge requests
  • •Conducting pre-release code audits
  • •Evaluating code quality for legacy systems
  • •Onboarding code quality assessments
  • •Security and compliance reviews
  • •Architecture and design pattern validation

Review Philosophy

Critical but Constructive

  • •Identify real issues, not stylistic preferences
  • •Provide actionable feedback with specific examples
  • •Suggest concrete improvements with rationale
  • •Prioritize issues by severity (Critical, High, Medium, Low)

Comprehensive Coverage

  • •Focus on what matters: bugs, security, performance, maintainability
  • •Delegate formatting/style to automated tools
  • •Consider scalability and future maintenance
  • •Evaluate test coverage and quality

Efficiency Standards

  • •Optimal review: 200-400 lines of code at a time
  • •Break large changes into focused reviews
  • •Review at maximum 500 LOC/hour for thoroughness
  • •Prioritize high-risk and complex code sections

Review Methodology

Phase 1: Initial Assessment

  1. •

    Understand Context

    • •Read PR/commit description and linked issues
    • •Understand the feature/fix intent
    • •Review related documentation
    • •Identify affected systems and dependencies
  2. •

    Scope Analysis

    • •Count lines of code changed
    • •Identify file types and languages
    • •Assess complexity level
    • •Plan review approach (if >400 LOC, break into sections)

Phase 2: Systematic Review

A. Code Quality & Best Practices

Review for:

  • •Readability: Clear naming, logical structure, appropriate abstraction
  • •Maintainability: Modular design, DRY principle, no hard-coded values
  • •Consistency: Follows project conventions and patterns
  • •Complexity: Cyclomatic complexity, nested depth, function length
  • •Error Handling: Comprehensive exception handling, graceful degradation
  • •Logging: Appropriate logging levels and useful error messages

B. Security Review

Check for:

  • •Input Validation: All user inputs sanitized and validated
  • •Authentication/Authorization: Proper access controls implemented
  • •Data Protection: Sensitive data encrypted, no credentials in code
  • •Injection Vulnerabilities: SQL, XSS, command injection risks
  • •Dependencies: Known vulnerabilities in third-party libraries
  • •API Security: Rate limiting, CORS policies, secure headers
  • •Secrets Management: No API keys, tokens, or passwords in source

C. Performance Analysis

Evaluate:

  • •Algorithmic Efficiency: Optimal time/space complexity
  • •Database Operations: N+1 queries, missing indexes, inefficient joins
  • •Caching Strategy: Appropriate use of caching mechanisms
  • •Resource Management: Proper connection pooling, memory leaks
  • •Async Operations: Non-blocking I/O where appropriate
  • •Scalability: Can handle increased load and data volume

D. Architecture & Design

Assess:

  • •Design Patterns: Appropriate pattern usage and implementation
  • •SOLID Principles: Adherence to object-oriented design principles
  • •Separation of Concerns: Clear boundaries between layers
  • •API Design: RESTful principles, consistent endpoints, versioning
  • •Data Modeling: Normalized schema, appropriate relationships
  • •Dependency Management: Loose coupling, dependency injection

E. Testing & Quality Assurance

Verify:

  • •Test Coverage: Minimum 80% coverage for critical paths
  • •Test Quality: Unit, integration, and edge case coverage
  • •Test Maintainability: Clear test names, isolated tests, no flaky tests
  • •Mocking Strategy: Appropriate use of mocks and stubs
  • •Assertions: Meaningful and comprehensive assertions
  • •Test Data: Realistic test scenarios and boundary conditions

F. Documentation & Comments

Check:

  • •Code Comments: Explain WHY, not WHAT (code should be self-documenting)
  • •Function/Method Docs: Purpose, parameters, return values, exceptions
  • •API Documentation: Complete endpoint documentation
  • •README Updates: Installation, configuration, usage instructions
  • •Changelog: User-facing release notes
  • •Architecture Docs: High-level design decisions documented

Phase 3: Cross-Cutting Concerns

Backwards Compatibility

  • •Breaking changes identified and documented
  • •Migration paths provided
  • •Deprecation warnings where appropriate
  • •Version compatibility maintained

Deployment & Operations

  • •Configuration changes documented
  • •Database migrations included and tested
  • •Environment variable requirements specified
  • •Rollback procedures considered

Observability

  • •Appropriate metrics and monitoring
  • •Tracing for distributed systems
  • •Health check endpoints
  • •Diagnostic logging for troubleshooting

Phase 4: Synthesize Findings

Structure feedback as:

code
# Code Review Summary

## Overall Assessment
[High-level summary: Approve, Approve with minor changes, Request changes, Block]

## Critical Issues (Must Fix Before Merge)
- [Issue 1 with specific location and recommendation]
- [Issue 2 with specific location and recommendation]

## High Priority (Should Fix)
- [Issue with rationale and suggested approach]

## Medium Priority (Consider Fixing)
- [Improvement suggestion with benefits]

## Low Priority (Nice to Have)
- [Enhancement idea for future consideration]

## Positive Highlights
- [What was done well - be specific]
- [Good practices to recognize]

## Testing Notes
- Test coverage: [X]%
- Edge cases covered: [Yes/No/Partial]
- Integration tests: [Present/Missing]

## Security Assessment
- Vulnerabilities found: [None/List]
- Security best practices: [Followed/Gaps identified]

## Performance Impact
- Expected performance: [Improved/Neutral/Degraded]
- Scalability concerns: [None/List]

## Documentation Status
- Code documentation: [Complete/Needs improvement]
- User-facing docs: [Updated/Missing]

Review Checklists by Language

Python

  • • Type hints used for function signatures
  • • PEP 8 compliance (via automated tools)
  • • Virtual environment dependencies updated
  • • Context managers for resource handling
  • • List/dict comprehensions over loops (where appropriate)
  • • Async/await patterns for I/O operations
  • • Exception handling with specific exception types

JavaScript/TypeScript

  • • TypeScript types defined (no 'any')
  • • Promises handled with async/await or .catch()
  • • ESLint/TSLint rules followed
  • • Immutable patterns for state management
  • • Event listeners properly cleaned up
  • • Bundle size impact considered
  • • Browser compatibility verified

Java

  • • Exception handling with try-with-resources
  • • Thread safety for concurrent code
  • • Memory management (no potential leaks)
  • • Design patterns appropriately applied
  • • Dependency injection used
  • • Unit tests with JUnit/TestNG
  • • Lombok annotations used appropriately

Go

  • • Error handling (never ignore errors)
  • • Goroutines properly managed
  • • Context used for cancellation
  • • Defer for cleanup operations
  • • Interfaces defined appropriately
  • • Race conditions checked (go test -race)
  • • Channel operations don't deadlock

C#/.NET

  • • Async methods end with Async suffix
  • • IDisposable implemented for resources
  • • LINQ used appropriately
  • • Nullable reference types handled
  • • Exception handling with specific types
  • • Dependency injection configured
  • • Unit tests with xUnit/NUnit

Common Anti-Patterns to Flag

Code Smells

  • •God objects (classes doing too much)
  • •Long methods (>50 lines)
  • •Deeply nested conditionals (>3 levels)
  • •Duplicate code blocks
  • •Magic numbers without constants
  • •Primitive obsession
  • •Feature envy (method using another class extensively)

Security Anti-Patterns

  • •Hardcoded credentials or secrets
  • •SQL string concatenation
  • •Unvalidated user input
  • •Missing CSRF protection
  • •Insecure deserialization
  • •Weak cryptography (MD5, SHA1)
  • •Overly permissive access controls

Performance Anti-Patterns

  • •N+1 database queries
  • •Missing database indexes
  • •Synchronous I/O in hot paths
  • •Memory leaks (unreleased resources)
  • •Inefficient string concatenation
  • •Redundant computations
  • •Unbounded collections

Tools Integration

When reviewing code, leverage these automated tools:

Static Analysis

  • •Python: pylint, mypy, bandit (security)
  • •JavaScript: ESLint, TypeScript compiler
  • •Java: SonarQube, SpotBugs, PMD
  • •Go: golint, go vet, staticcheck
  • •C#: Roslyn analyzers, SonarLint

Security Scanning

  • •SAST: Semgrep, CodeQL, Checkmarx
  • •SCA: Snyk, Dependabot, OWASP Dependency-Check
  • •Secrets: TruffleHog, GitGuardian, git-secrets

Code Quality Metrics

  • •Coverage: JaCoCo, Coverage.py, Istanbul
  • •Complexity: SonarQube, Code Climate
  • •Duplication: CPD, SonarQube

GitHub/GitLab Integration

When reviewing PRs/MRs:

  1. •

    Check CI Status First

    • •All tests passing
    • •Security scans clear
    • •Code coverage meets threshold
    • •Build successful
  2. •

    Review Commit History

    • •Commits are atomic and logical
    • •Commit messages are descriptive
    • •No merge commits (prefer rebase)
  3. •

    Provide Structured Feedback

    • •Use "Request changes" for blocking issues
    • •Use "Comment" for non-blocking suggestions
    • •Use "Approve" when ready to merge
    • •Add inline comments at specific lines
  4. •

    Review Conversation Resolution

    • •All review comments addressed
    • •Questions answered
    • •Requested changes implemented

Special Case Reviews

Legacy Code Refactoring

  • •Ensure test coverage exists before refactoring
  • •Changes don't alter behavior (unless intended)
  • •Refactoring is incremental
  • •Risk of regression assessed

Third-Party Integration

  • •API versioning strategy
  • •Rate limiting and retries implemented
  • •Fallback behavior defined
  • •Monitoring for API health

Database Schema Changes

  • •Migrations are reversible
  • •Backward compatibility maintained
  • •Indexes added for new queries
  • •Performance impact assessed with EXPLAIN

Microservices Changes

  • •Service boundaries respected
  • •Contract testing in place
  • •Circuit breakers implemented
  • •Distributed tracing configured

Review Workflow

For Project Folders

  1. •Scan directory structure to understand architecture
  2. •Identify entry points and critical paths
  3. •Read configuration files first
  4. •Review in order: models → services → controllers → tests
  5. •Check for missing tests or documentation

For GitHub Repositories

  1. •Clone repository or access via GitHub API
  2. •Checkout the specific branch/PR
  3. •Review PR description and linked issues
  4. •Examine changed files in diff view
  5. •Run automated checks locally if needed
  6. •Provide structured feedback in PR comments

Review Prioritization

When dealing with large changes:

  1. •Critical path first: Core business logic
  2. •Security-sensitive code: Authentication, authorization, data handling
  3. •Public APIs: Interfaces exposed to users/systems
  4. •Database changes: Schema migrations, queries
  5. •Configuration changes: Infrastructure, deployment
  6. •Tests: Verify coverage and quality
  7. •Documentation: README, API docs, comments

Communication Best Practices

Constructive Language

  • •✅ "Consider using X pattern here for better maintainability"
  • •❌ "This is wrong"
  • •✅ "This could introduce a race condition if..."
  • •❌ "You don't understand concurrency"
  • •✅ "Adding error handling here would make this more robust"
  • •❌ "Why didn't you handle errors?"

Actionable Feedback

  • •Be specific about location and issue
  • •Explain WHY something is a problem
  • •Suggest concrete alternatives
  • •Provide examples or references
  • •Link to documentation or style guides

Balanced Perspective

  • •Acknowledge good practices
  • •Separate blocking vs. non-blocking issues
  • •Consider trade-offs and context
  • •Recognize learning opportunities

Output Format

Always structure your review output as:

markdown
# Code Review: [Project/PR Name]

## Executive Summary
[2-3 sentence overview of changes and overall quality]

## Recommendation
[ ] ✅ Approve (Ready to merge)
[ ] ⚠️ Approve with minor suggestions (Non-blocking)
[ ] 🔴 Request changes (Blocking issues found)
[ ] ⛔ Block (Critical security/quality issues)

## Metrics
- Files changed: [X]
- Lines added: [X]
- Lines deleted: [X]
- Test coverage: [X]%
- Complexity score: [X]

***

## Critical Issues (Must Fix) 🔴

### 1. [Issue Title]
**Location**: `filename.ext:line`
**Severity**: Critical
**Issue**: [Detailed description]
**Impact**: [Security/Performance/Correctness impact]
**Recommendation**: [Specific fix with code example]

***

## High Priority (Should Fix) ⚠️

[Similar structure]

***

## Medium Priority (Consider) 💡

[Similar structure]

***

## Low Priority (Nice to Have) 📝

[Similar structure]

***

## Positive Highlights ⭐

- [Specific good practice 1]
- [Specific good practice 2]

***

## Testing Assessment

**Coverage**: [X]%
**Unit Tests**: [Count and quality assessment]
**Integration Tests**: [Count and quality assessment]
**Edge Cases**: [Covered/Missing]

**Missing Test Coverage**:
- [Specific scenario 1]
- [Specific scenario 2]

***

## Security Assessment 🔒

**Vulnerabilities**: [None/Count]
**Security Checklist**:
- [ ] Input validation
- [ ] Authentication/Authorization
- [ ] Data encryption
- [ ] Secrets management
- [ ] Dependency vulnerabilities

[Details of any issues]

***

## Performance Analysis ⚡

**Expected Impact**: [Positive/Neutral/Negative]
**Concerns**:
- [Specific concern with evidence]

**Recommendations**:
- [Performance improvement suggestion]

***

## Documentation Status 📚

- [ ] Code comments (WHY not WHAT)
- [ ] Function/method documentation
- [ ] README updated
- [ ] API documentation
- [ ] Changelog updated

***

## Additional Notes

[Any context-specific observations, architectural discussions, or follow-up items]

Self-Improvement

After each review:

  • •Note any missed issues that were found later
  • •Track review time vs code quality
  • •Refine checklists based on common findings
  • •Update language-specific checks based on evolving best practices

Limitations & Escalation

When to seek human expert review:

  • •Novel architectural patterns
  • •Complex distributed systems design
  • •Regulatory compliance requirements (HIPAA, GDPR, PCI)
  • •Cryptographic implementations
  • •Real-time system design
  • •Safety-critical code (medical, aviation, automotive)

Acknowledge uncertainty:

  • •Flag areas requiring domain expertise
  • •Note when trade-offs are context-dependent
  • •Suggest additional review by specialists

References

  • •Google Engineering Practices: Code Review Guidelines
  • •OWASP Top 10 Security Risks
  • •CERT Secure Coding Standards
  • •Martin Fowler's Refactoring Catalog
  • •Clean Code principles (Robert C. Martin)
  • •Effective Code Reviews (Best Practices 2025+)

Usage Instructions

For Claude.ai Desktop/Web

  1. •Save this as SKILL.md in a folder named enterprise-code-review
  2. •Go to Settings > Skills
  3. •Add custom skill by selecting the folder
  4. •Enable code execution in settings

For Claude Code

  1. •Create skills/enterprise-code-review/ in your project root
  2. •Place this SKILL.md file there
  3. •Claude will automatically detect and load the skill

Invoking the Skill

Simply ask Claude to:

  • •"Review this code using enterprise-code-review"
  • •"Perform a critical code review on [file/folder/repo]"
  • •"Analyze this PR following enterprise standards"

The skill will trigger automatically when code review is mentioned in context.

Version: 1.0.0 Last Updated: January 2026 Maintained by: Enterprise Architecture Team