mcp-uxhints

mcp-uxhints

Provides MCP tools to browse, search, retrieve, and refresh bite-sized UX and product-design hints from the uxhints.com catalog over stdio, with local caching for offline access.

Category
访问服务器

README

mcp-uxhints

MCP (Model Context Protocol) server that exposes the uxhints.com catalog of bite-sized UX and product-design hints as MCP tools over stdio. Content comes from the site's public WordPress REST API (converted to clean Markdown), served through a local cache so tool calls are instant and work offline.

Install with npx (recommended)

Prerequisite: Node.js 18 or newer. No build step — npx downloads and runs the published package, and the bundled catalog snapshot lets the first tool call work with zero network.

Register the server in your MCP client, then restart the client.

OpenCode

Add to opencode.json:

{
  "mcp": {
    "mcp-uxhints": {
      "type": "local",
      "command": ["npx", "-y", "mcp-uxhints"],
      "enabled": true
    }
  }
}

Claude Desktop

Add to claude_desktop_config.json:

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

Note: do not run npx mcp-uxhints from inside your own clone of this repository — npm resolves the local package.json and fails with command not found. Any other directory works.

Install from source (alternative)

For development or pinning a local build:

git clone https://github.com/Mateodioev/mcp-uxhints.git
cd mcp-uxhints
npm install && npm run build

Then point your client at the built file instead of npx:

  • OpenCode: "command": ["node", "/absolute/path/mcp-uxhints/build/index.js"]
  • Claude Desktop: "command": "node", "args": ["/absolute/path/mcp-uxhints/build/index.js"]

Tools

Tool What it does
list_categories List all categories with slugs and hint counts.
list_hints Paginated hint list, optionally filtered by category slug.
search_hints Local search over title, excerpt and content; case- and diacritic-insensitive, title matches first.
get_hint One hint by id or slug (exactly one), full content as Markdown.
random_hint One random hint, full content as Markdown.
refresh_hint_catalog Force a live sync and rewrite the local cache.

Typical flow: list_categorieslist_hintsget_hint. Slugs never include the category prefix (occams-razor-law-in-ux-design, not ux-laws/occams-razor-law-in-ux-design).

Caching

Tool calls read from a local catalog cache (default TTL 24h) instead of hitting the API every time. Resolution order:

Layer Location Role
1. User cache OS cache dir, catalog.json Read first; refreshed when older than the TTL.
2. Bundled snapshot catalog.snapshot.json inside the package Ships fresh at publish time; seeds the user cache with zero network.
3. Live API uxhints.com Source of truth for refreshes; if it fails, stale caches are served with a notice in the output.

Not fresh? Call refresh_hint_catalog, or wait for the TTL to expire. A postinstall hook warms the user cache automatically when the bundled snapshot has expired since publish.

Env var Default Effect
UXHINTS_CACHE_TTL_HOURS 24 Hours a cached catalog counts as fresh (float; invalid values → 24).
UXHINTS_CACHE_DIR OS cache dir¹ Override where catalog.json lives.
UXHINTS_NO_CACHE unset Set to 1 to bypass all caching and call the live API on every tool call.

¹ ~/.cache/mcp-uxhints on Linux (honors XDG_CACHE_HOME), ~/Library/Caches/mcp-uxhints on macOS, %LOCALAPPDATA%\mcp-uxhints\Cache on Windows.

Details

Topic Notes
Data source Public WordPress REST API of uxhints.com (/wp-json/wp/v2)
Content format WordPress HTML → Markdown at sync time (block comments stripped, entities decoded)
Caching 3 layers: user cache → bundled snapshot → live API (see above)
Resilience 15 s request timeout; live failures fall back to stale caches; all failures surface as MCP tool errors, never crash the server
Scripts npm run build (tsc), npm run sync:snapshot (regenerate the bundled snapshot), npm start (run the built server)

Credit

All content © uxhints.com by Paul Capcan. This server only wraps the site's public API; it does not host or modify the content.

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