# Agentic Architecture

> Enforces high-level architectural thinking, separation of concerns, and scalability checks before coding.

- Skill: `diegosouzapw/agentic-architecture` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add diegosouzapw/agentic-architecture`
- Raw SKILL.md: https://api.skillmd.com/api/skills/diegosouzapw/agentic-architecture/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: diegosouzapw (https://skillmd.com/u/diegosouzapw)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/diegosouzapw/agentic-architecture

---


# Agentic Architecture Protocol

## 1. Think Before You Code
Before implementing any feature that spans multiple files:
1.  **Analyze Data Flow**: Where does data come from? Where does it go?
2.  **Define Interfaces**: creating `types/*.ts` is often the best first step.
3.  **Check Boundaries**: Ensure API logic stays in `api/`, UI in `components/`, and business logic in `services/` or `hooks/`.

## 2. Scalability & Performance Checks
- **Database**:
    - Are we fetching 1000 items to filter 10? (Use DB filters instead).
    - Is RLS (Row Level Security) compatible with this query?
- **Frontend**:
    - Are we causing unnecessary re-renders? (Use `React.memo`, `useCallback` appropriately).
    - Is this component becoming a "God Component"? (Break it down).

## 3. The "Three-Tier" Rule
For any non-trivial feature, verify you have these three layers:
1.  **Data Layer**: Types + API/Service (e.g., `user.types.ts`, `userService.ts`)
2.  **State Layer**: Hook or Store (e.g., `useUser.ts`)
3.  **View Layer**: Components (e.g., `UserProfile.tsx`)

## 4. Architecture Checklist
- [ ] Have I defined the types first?
- [ ] Is the business logic separated from the UI?
- [ ] Did I consider how this scales to 10,000 users/items?
- [ ] Is the database schema validated (if changing DB)?