mcp-software-design

mcp-software-design

Provides software design guidance via MCP, including SOLID principles, design patterns, pattern scaffolding, and heuristic code-smell detection to help review and refactor code.

Category
访问服务器

README

mcp-software-design

npm version npm downloads License: MIT MCP

An MCP server that teaches and helps apply software-design guidance — the SOLID principles, the OOP pillars, DRY / KISS / YAGNI / meaningful naming / clean code, and the 23 Gang-of-Four design patterns — plus pattern scaffolding and heuristic code-smell detection.

It's the companion to mcp-udacity-commit: same stack (TypeScript, the MCP SDK, stdio transport), same shape (pure logic modules + thin server wiring).

Install

Register it with Claude Code — one line, nothing to clone:

claude mcp add software-design -- npx -y mcp-software-design

Or in an MCP client config:

{
  "mcpServers": {
    "software-design": {
      "command": "npx",
      "args": ["-y", "mcp-software-design"]
    }
  }
}

Why this exists — and its one honest caveat

The commit server can lint: "subject ≤ 50 chars" is objectively checkable. Design principles and patterns are not like that — "does this violate SRP?" or "should this be a Factory?" are judgment calls, not lint rules.

So this server does not pretend to grade your architecture pass/fail. Instead it does the parts that are genuinely reliable, and hands the judgment to the model:

Capability Primitive What it gives you
Reference resources + explain_concept Authoritative, consistent definitions so the model cites the same thing every time.
Scaffolding scaffold_pattern A language-agnostic skeleton of a pattern's participants.
Smell heuristics check_smells A few genuinely-checkable proxies (long method, deep nesting, …) — hints, never verdicts.
Explain / apply review_design, apply_pattern prompts Prime the model to review or refactor, grounded in the tools + resources above.

Design analysis is a judgment call, so the "explain/apply helper" is exposed as MCP prompts (which drive the client's model) rather than server code pretending to understand your snippet.

Tools

  • list_catalog { kind? } — list concepts, optionally filtered (principle | solid | oop | pattern | creational | structural | behavioral). solid / oop narrow to the SOLID five / the four OOP pillars.
  • explain_concept { name } — full guidance for one principle or pattern (intent, when-to-use, trade-offs, participants). Accepts a slug, name, or alias ("SRP", "open-closed", "pubsub").
  • scaffold_pattern { pattern, names? } — pseudo-code skeleton for a GoF pattern; names optionally renames roles to your domain ({ "Product": "Notification" }).
  • check_smells { code, …thresholds? } — heuristic scan for long method, large class, long parameter list, deep nesting, duplication, and large file. Each finding names the principle it hints at plus a suggested refactor. All thresholds are overridable per call.

Resources

  • design://principles — SOLID, OOP pillars, DRY, KISS, YAGNI, meaningful naming, and more.
  • design://patterns — the 23 GoF patterns, grouped creational / structural / behavioral.
  • design://smells — what check_smells detects, its thresholds, and its caveats.

Prompts

  • review_design { code, focus? } — review a snippet against the principles/patterns, grounded in check_smells + the resources.
  • apply_pattern { pattern, code } — refactor a snippet to apply a named pattern (and first judge whether it even fits).

Build from source

For local development, or to run a local checkout instead of the published package:

npm install
npm run build      # compiles src → build
npm start          # runs the stdio server
npm test           # builds, then runs the unit tests
npm run test:client  # end-to-end check against the built server

Then register it the same way as Install above — both the claude mcp add command and the MCP-client-config form work — but point at your local build instead of npx:

claude mcp add software-design -- node /absolute/path/to/mcp-software-design/build/index.js

Layout

src/
  catalog.ts   # principles + 23 GoF patterns (data + lookup + markdown)
  smells.ts    # pure, testable code-smell heuristics
  scaffold.ts  # renders a pattern's participants into a skeleton
  index.ts     # MCP wiring: resources, tools, prompts
test/
  catalog.test.mjs   # catalog lookup + scaffolder
  smells.test.mjs    # smell detectors (incl. string/comment edge cases)

The src/*.ts logic modules are pure and side-effect-free, so they're unit tested directly against the compiled output — the server (index.ts) is only thin wiring on top.

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选