alyanhaider/system-architecture

🧠 Production-Grade AI Coding Skills & Rules

A curated collection of production-grade skill files and rule systems designed for AI coding agents (Claude 3.5 Sonnet, Cursor, Windsurf, Claude Code, and GitHub Copilot).

---

⚡ The Core Problem

AI agents write code that works now, not code that scales in 3 weeks.

When prompting an AI to build a feature, it defaults to the path of least resistance:

• It hardcodes CORS origins to * (security vulnerability).
• It dumps business logic straight into Express routes or controllers.
• It writes database queries inside loops (generating N+1 query performance kills).
• It stores session tokens in browser localStorage (making sessions stealable via XSS).

This repository bridges the gap. By loading these specific skill files into your AI workspace, you enforce strict architectural boundaries, industry-standard security patterns, and robust coding rules automatically—without having to repeat yourself in every prompt session.

---

📁 Skills Index

| Skill Domain | Description | Prevents these AI Blunders | |---|---|---| | 🔒 authentication-security | secure cookie configurations, JWT access/refresh rotation, password hashing, OAuth state | Storing tokens in localStorage, MD5/SHA256 password storage, missing OAuth state parameter | | 🛡️ security-hardening | input sanitization, rate-limiting rules, security headers (Helmet), injection prevention | SQL injection, prototype pollution, missing request limits, raw logging of passwords | | 🌐 [**