claude-ai-mcp-demo

claude-ai-mcp-demo

MCP server that enables Claude to dynamically rebrand and restock a themeable micro-storefront, including content and look-and-feel control via MCP tools.

Category
访问服务器

README

claude ai mcp demo

Pop-up shop in a box. A themeable micro-storefront that Claude can rebrand and restock end-to-end through MCP — same site, swappable identity. "Make this an artisanal coffee roaster" → palette, fonts, hero, and product list all flip live. Then "actually, make it a vintage record store" → it flips again.

Built to show, at internal demos like Digital AI Sharing, that an MCP server can drive a website's content AND its look-and-feel — not just CRUD a row in a CMS.

What it shows

  • Content control — Products collection exposed over the official @payloadcms/plugin-mcp, plus bulk tools (createManyProducts, updateManyProducts) with per-item transactions and partial success.
  • Look-and-feel control (planned, see Status)SiteSettings for theme tokens (palette, typography, brand) and Pages/Sections for hero/featured/footer blocks — all writable over MCP.
  • Safety rails — every MCP write lands as a draft (forceDraftForMcpWrites hook) so a human can review before publish. MCP identity is a non-admin mcp-agents row, not a human admin.
  • Two connection paths — paste a static API key (Claude Code, mcp-remote) OR connect via OAuth from Claude.ai web (WorkOS AuthKit as the authorization server).

Status

Capability State
products collection over MCP ✅ working
Bulk create/update tools with draft-on-write ✅ working
mcp-agents non-admin identity + per-key tool gating ✅ working
Themeable storefront frontend 🚧 placeholder page only
SiteSettings theme tokens 🚧 not yet
Pages / Sections block library 🚧 not yet
WorkOS OAuth for Claude.ai web 🚧 env scaffolded, wiring WIP

The themeable frontend and the site-config collections are the next chunk of work — that's the headline demo moment.

Setup

docker compose up -d                 # local Postgres on :5432
pnpm install
cp .env.example .env.local
# fill in PAYLOAD_SECRET — `openssl rand -hex 32`
# (WorkOS vars only needed for the Claude.ai web OAuth path)
pnpm dev

Open http://localhost:3000/admin, create the first admin, then:

  1. Create an mcp-agents row (admin → MCP Agents).
  2. Create an MCP API key (admin → MCP API Keys) bound to that agent.
  3. Tick the createManyProducts / updateManyProducts checkboxes on the API-key row (custom tools are off-by-default).
  4. Connect from an MCP client to http://localhost:3000/api/mcp with the API key as a bearer token.

Migrating from the older mcp-products-demo setup? The DB was renamed (mcp_products_democlaude_ai_mcp_demo). Run docker compose down -v to nuke the old volume, then docker compose up -d for a fresh DB.

The demo moment (target)

Once the themeable frontend lands, the showcase prompt is something like:

"Rebrand this site as Ember & Oat, a small-batch coffee roaster. Warm earth-tone palette, serif headings, friendlier copy. Replace the products with 5 single-origin beans, prices $18–24, with tasting-note descriptions."

Claude calls a handful of MCP tools (update SiteSettings, replace Sections, bulk-create Products), the dev server picks up the drafts on next reload, the site is a coffee shop. Re-prompt with a different niche → it's something else.

Stack

  • Next.js 16 (App Router) + React 19
  • Payload CMS 3.85 with Postgres adapter
  • @payloadcms/plugin-mcp — Streamable HTTP at /api/mcp
  • WorkOS AuthKit (optional) — OAuth authorization server for Claude.ai web connections
  • pnpm + Docker Compose for local Postgres

Layout

app/
  (frontend)/        # public site — themeable storefront (WIP)
  (payload)/         # admin UI + /api/mcp endpoint
collections/         # Admins, MCPAgents, Products (+ SiteSettings/Pages soon)
lib/
  mcp-tools.ts       # createManyProducts, updateManyProducts
  categories.ts
payload.config.ts    # plugin wiring, MCP_ENABLED kill switch

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