Kitchen MCP Server
Enables querying food nutritional information, discovering recipes by ingredients or diet type, getting ingredient substitutions, and receiving personalized food recommendations based on mood and season.
README
🥗 Kitchen MCP Server
<div align="center"> <b>Model Context Protocol server for food, nutrition, and smart recipe recommendations</b> </div>
A Model Context Protocol (MCP) server focused on food, nutrition, and recipe suggestions. This project allows you to query ingredients, get nutritional information, and receive recipe recommendations using the open MCP standard, making it compatible with Claude Desktop and other MCP clients.
Features
- Query foods and their nutritional information.
- Search for foods by nutritional criteria (protein, fat, calories).
- List available ingredients.
- Get all available recipes.
- Get recipe suggestions based on nutrition.
- Find recipes by ingredients.
- Suggest recipes by diet type (vegan, keto, etc.) and calories.
- Suggest substitutes for a given ingredient.
- Suggest necessary kitchen utensils for a recipe.
- Recommend foods or recipes based on your mood and optionally the season.
- All endpoints follow the MCP protocol and are compatible with Claude Desktop and other MCP clients.
Project Structure
src/mcp-server.js: Main MCP server implementation (Node.js).src/utils.js: Utility functions (e.g., Levenshtein distance).data/: Contains JSON and CSV files with food and recipe data.
Requirements
- Node.js 18 or higher
- npm
Installation
-
Clone the repository:
git clone https://github.com/paulabaal12/kitchen-mcp.git cd kitchen-mcp -
Install dependencies:
npm install -
Make sure your data files (
ingredientes_unificados.json,recetas_unificadas.json) are present in thesrc/data/directory.
Usage
Running the MCP Server
To start the server for use with Claude Desktop:
node src/mcp-server.js
Claude Desktop Configuration
Add the following to your claude_desktop_config.json:
{
"mcpServers": {
"kitchen": {
"command": "node",
"args": [
"D:/Documentos/GitHub/kitchen-mcp/src/mcp-server.js"
],
"env": {
"NODE_ENV": "production"
}
}
}
}
Adjust the path if your project is in a different location.
Example: Get Recipes by Ingredients
Prompt in Claude Desktop:
I want to cook something with apple, sugar and butter.
Sample MCP Tool Call (Direct Format):
{
"method": "get_recipes_by_ingredients",
"params": {
"ingredients": ["apple", "sugar", "butter"]
}
}
Sample MCP Tool Call (Generic Format):
{
"method": "tools/call",
"params": {
"name": "get_recipes_by_ingredients",
"arguments": {
"ingredients": ["apple", "sugar", "butter"]
}
}
}
Sample Response:
{
"content": [
{
"type": "text",
"text": "[\n {\n \"name\": \"Fruit Galette\",\n \"ingredients\": [\"apple\", \"sugar\", \"butter\", ...],\n ...\n },\n ...\n]"
}
]
}
Example Recipe: Creole Cream Cheesecake With Caramel-Apple Topping
Ingredients:
- Graham cracker crumbs, sugar, butter, cream cheese, eggs, apples, honey, cinnamon, cardamom, ginger, vanilla, lemon.
Preparation:
- Prepare the crust and bake.
- Mix and bake the cheesecake filling.
- Make the caramel-apple topping in a pan.
- Chill the cheesecake, then top with apples before serving.
Approximate Calories:
~7,700 calories for the whole cheesecake (about 480–640 per slice).
MCP Tools (Endpoints)
The following MCP tools are available:
| Name | Description |
|---|---|
get_foods |
Get all available foods. |
get_food_by_name |
Find a specific food by name. |
search_foods |
Search foods by nutritional criteria (minProtein, maxFat, maxCalories). |
get_ingredients |
Get list of available ingredients. |
get_recipe_suggestions |
Get recipe suggestions based on nutritional content. |
get_recipes |
Get all available recipes. |
get_recipes_by_ingredients |
Find recipes by specific ingredients. |
suggest_recipe_by_diet |
Suggest recipes by diet type (vegan, keto, etc). Optionally filter by calories. |
suggest_ingredient_substitution |
Suggest substitutes for a given ingredient (e.g., orange juice). |
suggest_utensils_for_recipe |
Suggest necessary kitchen utensils for a given recipe (by name). |
recommend_by_mood_and_season |
Recommends foods or recipes based on mood and optionally season (e.g., happy + summer). |
Example: Recommend by Mood and Season
Prompt:
I feel happy, what can I eat?
Sample MCP Tool Call (Direct Format):
{
"method": "recommend_by_mood_and_season",
"params": {
"mood": "happy"
}
}
Sample MCP Tool Call (Generic Format):
{
"method": "tools/call",
"params": {
"name": "recommend_by_mood_and_season",
"arguments": {
"mood": "happy"
}
}
}
Sample Response:
{
"content": [
{
"type": "text",
"text": "[\n {\"name\": \"Fruit Salad\", ...}, ...]"
}
]
}
- If you add a
season(e.g.,"season": "autumn"), the recommendations will also consider seasonal foods. - You can set
typetofoodorrecipe(default is recipe).
Example: Suggest Utensils for a Recipe
Prompt:
What utensils do I need for lasagna?
Sample MCP Tool Call (Direct Format):
{
"method": "suggest_utensils_for_recipe",
"params": {
"recipe_name": "lasagna"
}
}
Sample MCP Tool Call (Generic Format):
{
"method": "tools/call",
"params": {
"name": "suggest_utensils_for_recipe",
"arguments": {
"recipe_name": "lasagna"
}
}
}
Sample Response:
{
"content": [
{
"type": "text",
"text": "{\n \"recipe\": \"lasagna\",\n \"utensils\": [\"pan\", \"pot\", ...]\n}"
}
]
}
Visual Examples with Claude Desktop
Example: Recipe search by ingredients or diet
Recipe search using ingredients in Claude Desktop.
Example: Recommendation by mood
Food recommendation based on mood in Claude Desktop.
Example: Utensil suggestion
Automatic utensil suggestion for a recipe in Claude Desktop.
Example: Ingredient substitution
Ingredient substitution suggestion in Claude Desktop
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。