ai-generated-business-code-review_skill

Overview

This skill evaluates AI-generated business and application code for correctness, robustness, maintainability, performance, and security. It produces a numeric 0–10 score, a concise risk level, and a must-fix checklist with concrete evidence. For C++ code, it enforces OpenHarmony C++ and OpenHarmony security review constraints as mandatory checks.

How this skill works

I identify the language and whether the code is business logic or test code, and refuse test-code reviews in favor of a unit-test review workflow. I inspect behavior against requirements, edge cases, error handling, resource management, and security controls. For C++ code I apply openharmony-cpp and openharmony-security-review as hard constraints and flag any violations in evidence and must-fix items.

When to use it

• Review AI-generated backend, service, or application code for production readiness
• Need a numeric quality score plus a clear risk level and prioritized fixes
• Require explicit evidence lines or snippets to support findings
• Assess C++ business code for OpenHarmony coding and security compliance
• Validate security, resource management, and error-handling before deployment

Best practices

• Provide the codebase or specific files and relevant requirements or usage scenarios
• Include expected inputs, error modes, and performance constraints for accurate scoring
• Mark C++ reviews as such so OpenHarmony constraints are applied
• Expect at least two concrete evidence items; lack of evidence lowers score
• Request follow-up iterations after fixes to verify remediations

Example use cases

• Score a new AI-generated microservice implementation and get a must-fix checklist
• Audit C++ service code to ensure OpenHarmony C++ and security rules are met
• Compare two AI-generated implementations by normalized 0–10 rubric
• Identify missed edge cases, swallowed exceptions, and resource leaks before release
• Produce developer-facing fixes with file/line or snippet references