A modular library of high-authority Markdown configuration files designed to govern AI agents (like Cursor, GitHub Copilot, or GPT-4). This repository ensures that generated code adheres to strict architectural standards, performance benchmarks (Lighthouse 100), and specific framework philosophies.
| File | Description |
|---|---|
orchestrator.md |
The master prompt to initialize an AI session and link all rules. |
persona.md |
Defines the "Staff Engineer" persona and authority level. |
vue-rules.md |
Vue 3 Composition API standards (including AlexOp patterns). |
react-rules.md |
React 18+ standards for hooks, performance, and RTL testing. |
design-system.md |
UI/UX guidelines (8pt grid, A11y, Dark Mode, Lighthouse). |
api-testing-global.md |
Rules for Axios/Fetch, Vitest, and strict test assertions. |
When starting a new conversation with an AI agent, copy the contents of orchestrator.md and paste it as your first message. The AI will then ask for the relevant framework files based on your project.
If you use Cursor, you can copy the contents of these files into your .cursorrules file or reference this repository folder to provide constant context to the IDE.
- Strict Mocking: Mandatory use of
vi.hoisted()andtoHaveBeenCalledTimes(). - Isolation: Components are treated as black boxes; sub-components are always stubbed.
- Lighthouse 100: All generated components must prioritize semantic HTML, image optimization, and accessibility.
- A11y: Mandatory
aria-labelfor icon buttons and strict color contrast ratios.
- Domain Agnostic: Props are defined by UI role, not backend entities.
- Logic Extraction: Complex logic must live in Composables (Vue) or Custom Hooks (React).
This project is licensed under the MIT License - see the LICENSE file for details.