system-design

Anthropic ｜ 架構與重構 ｜ 原文 ｜ Apache-2.0

做系統設計：需求、高階設計、深入細節、規模與取捨

---
name: system-design
description: 設計系統、服務與架構。用「design a system for」「how should we architect」「system design for」「what's the right architecture for」觸發，或當使用者需要 API 設計、資料建模、服務邊界方面的協助時使用。
---

# 系統設計

協助設計系統，並評估架構上的決策。

## 框架

### 1. 蒐集需求
- 功能需求（它要做什麼）
- 非功能需求（規模、延遲、可用性、成本）
- 限制條件（團隊人數、時程、既有技術棧）

### 2. 高階設計
- 元件圖
- 資料流
- API 契約
- 儲存方案的選擇

### 3. 深入細節
- 資料模型設計
- API endpoint 設計（REST、GraphQL、gRPC）
- 快取策略
- Queue／事件設計
- 錯誤處理與重試邏輯

### 4. 規模與可靠性
- 負載估算
- 水平擴展 vs. 垂直擴展
- Failover 與備援
- 監控與告警

### 5. 取捨分析
- 每個決策都有取捨，要明講出來。
- 考慮：複雜度、成本、團隊熟悉度、上市時間、可維護性

## 輸出

產出清楚、有結構的設計文件，附上圖（ASCII 或文字描述）、明確的假設，以及取捨分析。一定要指出系統成長之後你會回頭重新檢視哪些地方。

---
name: system-design
description: Design systems, services, and architectures. Trigger with "design a system for", "how should we architect", "system design for", "what's the right architecture for", or when the user needs help with API design, data modeling, or service boundaries.
---

# System Design

Help design systems and evaluate architectural decisions.

## Framework

### 1. Requirements Gathering
- Functional requirements (what it does)
- Non-functional requirements (scale, latency, availability, cost)
- Constraints (team size, timeline, existing tech stack)

### 2. High-Level Design
- Component diagram
- Data flow
- API contracts
- Storage choices

### 3. Deep Dive
- Data model design
- API endpoint design (REST, GraphQL, gRPC)
- Caching strategy
- Queue/event design
- Error handling and retry logic

### 4. Scale and Reliability
- Load estimation
- Horizontal vs. vertical scaling
- Failover and redundancy
- Monitoring and alerting

### 5. Trade-off Analysis
- Every decision has trade-offs. Make them explicit.
- Consider: complexity, cost, team familiarity, time to market, maintainability

## Output

Produce clear, structured design documents with diagrams (ASCII or described), explicit assumptions, and trade-off analysis. Always identify what you'd revisit as the system grows.