Norish MCP Server
A Model Context Protocol server that connects AI assistants to the Norish recipe management API for recipe browsing, meal planning, and grocery list management.
README
Norish MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with access to the Norish recipe management API.
Connect your LLM client to your personal Norish instance for recipe browsing, meal planning, and grocery list management — all through a standardized MCP interface.
Features
This server exposes the following tools:
Recipes
| Tool | Description |
|---|---|
get_recipe |
Get a recipe by ID with ingredients and cooking steps |
list_recipes |
List recipes with optional search and pagination |
create_recipe |
Create a new recipe directly with structured data |
import_recipe_url |
Import a recipe from a URL (parses webpage) |
import_recipe_text |
Import a recipe from pasted text (plain text, JSON-LD, or YAML) |
Meal Planning
| Tool | Description |
|---|---|
get_today_planned_recipes |
Get planned recipes for today |
get_week_planned_recipes |
Get planned recipes for the current week |
get_month_planned_recipes |
Get planned recipes for the current month |
add_to_plan |
Add a recipe to your meal plan for a specific date and slot |
remove_from_plan |
Remove a planned recipe item from your meal plan |
Grocery List
| Tool | Description |
|---|---|
grocery_list |
List all items in the shopping list |
grocery_create |
Create a new grocery item |
grocery_mark_done |
Mark a grocery item as done (checked off) |
grocery_mark_undone |
Uncheck a previously checked-off grocery item |
grocery_delete |
Delete a grocery item from the shopping list |
grocery_assign_store |
Assign a grocery item to a specific store |
Stores
| Tool | Description |
|---|---|
list_stores |
List all stores |
create_store |
Create a new store with name, color, and icon |
System
| Tool | Description |
|---|---|
health_check |
Check the Norish API health (DB, parser service status) |
Prerequisites
- Node.js v22+ or Docker Compose
- A running Norish instance with API access
- Your Norish
NORISH_API_KEY(generate from your Norish account settings)
Setup
1. Clone and install dependencies
git clone <this-repo-url>
cd norish-mcp
npm install
2. Configure environment variables
Copy .env.example to .env and fill in the values:
cp .env.example .env
| Variable | Required | Description |
|---|---|---|
NORISH_API_URL |
Yes | The URL of your running Norish instance (e.g. https://meals.nashor.cloud:6002) |
NORISH_API_KEY |
Yes | Your API key from the Norish instance |
MCP_AUTH_ENABLED |
No (default: true) |
Set to "false" to disable authentication entirely. When enabled, MCP_API_KEY is required. |
MCP_API_KEY |
Yes (when auth enabled) | Secret key for authenticating requests to this MCP server. Clients must send it as a Bearer token in the Authorization header. |
ALLOWED_HOSTS |
No | Comma-separated list of allowed hostnames for DNS rebinding protection (e.g., "norish-mcp.my.domain"). Defaults to ['norish-mcp.my.domain']. |
PORT |
No (default: 3001) | Port the MCP server listens on |
3. Build and run
Development mode:
npm run dev
Production build:
npm run build
npm start
The server starts on http://localhost:<PORT> with the MCP endpoint at POST /mcp.
Docker (Recommended)
The easiest way to run this server is with Docker Compose using the official image:
# docker-compose.yml
services:
norish-mcp:
image: ghcr.io/kellojo/norish-mcp:latest
container_name: norish-mcp
ports:
- "3001:3001"
env_file: .env
Create a .env file in the same directory:
NORISH_API_URL=https://your-norish-instance.com:6002
NORISH_API_KEY=your-api-key-here
MCP_API_KEY=your-secret-mcp-key
PORT=3001
Then start it:
docker compose up -d
The server will be available at http://localhost:3001 with the MCP endpoint at POST /mcp.
Running Directly
Pull and run without Docker Compose:
docker pull ghcr.io/kellojo/norish-mcp:latest
docker run -d \
--name norish-mcp \
-p 3001:3001 \
--env-file .env \
ghcr.io/kellojo/norish-mcp:latest
Building from Source
If you prefer to build locally:
docker compose build
docker compose up -d
Authentication
By default, the server requires a Bearer token on every request:
Authorization: Bearer <MCP_API_KEY>
Requests without a valid token are rejected with a 401 Unauthorized error before reaching the MCP handler. This prevents unauthenticated access when exposing the server publicly.
To disable authentication (e.g., for local-only development), set MCP_AUTH_ENABLED=false in your .env. When disabled, no API key is required and clients can connect without an Authorization header.
Client Configuration
Claude Desktop (via mcp.json)
Add this to your Claude Desktop configuration (mcp.json):
{
"mcpServers": {
"norish": {
"command": "node",
"args": ["dist/index.js"],
"env": {
"NORISH_API_URL": "https://your-norish-instance.com:6002",
"NORISH_API_KEY": "<your-norish-api-key>",
"MCP_API_KEY": "<your-mcp-api-key>"
}
}
}
}
Other MCP Clients
For HTTP-based clients, point them to the server's endpoint with the auth header:
- URL:
http://localhost:<PORT>/mcp - Method:
POST - Headers:
Authorization: Bearer <MCP_API_KEY>,Content-Type: application/json
License
MIT
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。