Mealie MCP Server
Connects MCP clients like Claude to your Mealie recipe manager, enabling natural language search, creation, import, and updates of recipes.
README
Mealie MCP Server
An MCP (Model Context Protocol) server that connects Claude and other MCP clients to your Mealie recipe manager instance. Search, create, import, and update recipes through natural language.
Features
- Get Recipe — Retrieve full recipe details by slug or ID
- Search Recipes — Free-text search with filtering by category, tag, food, and tool
- Create Recipe — Create recipes from structured data (name, ingredients, instructions, etc.)
- Import Recipe — Import recipes by URL (Mealie scrapes the page automatically)
- Update Recipe — Modify any field on an existing recipe
Prerequisites
- Mealie v1+ instance running and accessible
- A Mealie API token (generate at
<your-mealie-url>/user/profile/api-tokens) - Node.js 20+ (for local usage) or Docker
Transports
The server supports two transports:
| Transport | Use case | Entry point |
|---|---|---|
| stdio (default) | Claude Desktop, Claude Code, local MCP clients | build/index.js |
| Streamable HTTP | Claude.ai web/mobile, remote access, multi-user | build/http.js |
Authentication
There are two separate auth layers — one for the server to talk to Mealie, and one for clients to talk to the server:
Claude.ai ──OAuth 2.0──▶ MCP Server ──API Token──▶ Mealie
-
Mealie API Token (
MEALIE_API_TOKEN) — The server's backend credential for accessing Mealie's REST API. Required for both transports. Generate one at<your-mealie-url>/user/profile/api-tokens. -
OAuth 2.0 + PKCE (HTTP transport only) — How remote clients like Claude.ai authenticate to the MCP server. Users log in with their Mealie username and password via a browser-based flow. The server validates credentials against Mealie and issues OAuth tokens. Not used with stdio transport.
In short: the API token lets the server access recipes, and OAuth lets users prove who they are.
Configuration
Copy .env.example to .env and fill in your values:
cp .env.example .env
Environment variables
| Variable | Required | Description |
|---|---|---|
MEALIE_URL |
Yes | Base URL of your Mealie instance (e.g. http://localhost:9925) |
MEALIE_API_TOKEN |
Yes | Your Mealie API token |
MCP_SERVER_URL |
HTTP mode only | Public URL of this server (e.g. https://mcp.your-domain.com) — used as the OAuth issuer URL |
PORT |
No | HTTP server port (default: 3000, HTTP mode only) |
TRANSPORT |
No | Set to http for HTTP mode in Docker (default: stdio) |
Usage
With Claude Desktop (stdio)
Add to your claude_desktop_config.json:
Using Docker:
{
"mcpServers": {
"mealie": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-e", "MEALIE_URL",
"-e", "MEALIE_API_TOKEN",
"mealie-mcp-server"
],
"env": {
"MEALIE_URL": "http://your-mealie-instance:9925",
"MEALIE_API_TOKEN": "your-api-token"
}
}
}
}
Using Node.js directly:
{
"mcpServers": {
"mealie": {
"command": "node",
"args": ["/path/to/mealie-mcp-server/build/index.js"],
"env": {
"MEALIE_URL": "http://your-mealie-instance:9925",
"MEALIE_API_TOKEN": "your-api-token"
}
}
}
}
With Claude Code (stdio)
claude mcp add mealie -- node /path/to/mealie-mcp-server/build/index.js \
-e MEALIE_URL=http://your-mealie-instance:9925 \
-e MEALIE_API_TOKEN=your-api-token
HTTP Transport (Claude.ai web / remote access)
The HTTP transport uses OAuth 2.0 with PKCE, with Mealie credentials as the identity provider. Users authenticate via a login page using their Mealie username and password. This makes it compatible with Claude.ai (web and mobile), which requires OAuth-capable MCP servers.
Requirements: The server must be reachable at a public HTTPS URL (MCP_SERVER_URL).
Using Node.js:
MEALIE_URL=http://your-mealie:9925 \
MEALIE_API_TOKEN=your-api-token \
MCP_SERVER_URL=https://mcp.your-domain.com \
node build/http.js
Using Docker:
docker run --rm \
-p 3000:3000 \
-e TRANSPORT=http \
-e MEALIE_URL=http://your-mealie:9925 \
-e MEALIE_API_TOKEN=your-api-token \
-e MCP_SERVER_URL=https://mcp.your-domain.com \
mealie-mcp-server
Using Docker Compose:
# Set env vars in .env file, then:
docker compose up mealie-mcp-server-http
Connecting Claude.ai to the HTTP endpoint:
In Claude.ai settings, add a new MCP server pointing at https://mcp.your-domain.com/mcp. Claude.ai will discover the OAuth metadata automatically and prompt you to log in with your Mealie credentials.
Nginx configuration:
See nginx-mcp.conf.example for the location blocks needed to proxy MCP and OAuth traffic when running behind nginx.
A health check endpoint is available at GET /health.
Note on auth state: The HTTP server stores OAuth tokens in memory. Restarting the server invalidates all existing sessions and requires clients to re-authenticate.
Tools
get_recipe
Get a recipe by its slug or ID.
| Parameter | Type | Required | Description |
|---|---|---|---|
slug |
string | Yes | Recipe slug or UUID |
search_recipes
Search and list recipes with optional filters.
| Parameter | Type | Required | Description |
|---|---|---|---|
search |
string | No | Free-text search query |
categories |
string[] | No | Filter by category slugs or IDs |
tags |
string[] | No | Filter by tag slugs or IDs |
foods |
string[] | No | Filter by food slugs or IDs |
tools |
string[] | No | Filter by tool slugs or IDs |
requireAllCategories |
boolean | No | Require all specified categories |
requireAllTags |
boolean | No | Require all specified tags |
page |
number | No | Page number (default: 1) |
perPage |
number | No | Results per page (default: 50) |
orderBy |
string | No | Sort field (e.g. createdAt, name) |
orderDirection |
asc | desc |
No | Sort direction |
create_recipe
Create a recipe from structured data.
| Parameter | Type | Required | Description |
|---|---|---|---|
name |
string | Yes | Recipe name |
description |
string | No | Short description |
recipeYield |
string | No | Yield (e.g. "4 servings") |
prepTime |
string | No | Prep time |
cookTime |
string | No | Cook time |
totalTime |
string | No | Total time |
recipeIngredient |
object[] | No | Ingredients list |
recipeInstructions |
object[] | No | Instruction steps |
nutrition |
object | No | Nutritional info |
recipeCategory |
object[] | No | Categories |
tags |
object[] | No | Tags |
notes |
object[] | No | Recipe notes |
import_recipe
Import a recipe by URL. Mealie scrapes the page and extracts recipe data.
| Parameter | Type | Required | Description |
|---|---|---|---|
url |
string | Yes | URL of the recipe page |
includeTags |
boolean | No | Extract tags from the page |
includeCategories |
boolean | No | Extract categories from the page |
update_recipe
Update fields on an existing recipe. Only provided fields are changed.
| Parameter | Type | Required | Description |
|---|---|---|---|
slug |
string | Yes | Recipe slug or UUID to update |
name |
string | No | New name |
description |
string | No | New description |
recipeYield |
string | No | New yield |
prepTime |
string | No | New prep time |
cookTime |
string | No | New cook time |
recipeIngredient |
object[] | No | Replacement ingredients |
recipeInstructions |
object[] | No | Replacement instructions |
nutrition |
object | No | Nutritional info |
recipeCategory |
object[] | No | Categories |
tags |
object[] | No | Tags |
notes |
object[] | No | Notes |
Development
Setup
npm install
Build
npm run build
Run in development
# stdio transport
MEALIE_URL=http://localhost:9925 MEALIE_API_TOKEN=your-token npm run dev
# HTTP transport
MEALIE_URL=http://localhost:9925 MEALIE_API_TOKEN=your-token \
MCP_SERVER_URL=http://localhost:3000 npm run dev:http
Test
npm test
# With coverage
npm run test:coverage
# Watch mode
npm run test:watch
Docker
# Build the image
docker build -t mealie-mcp-server .
# Run (stdio)
docker run -i --rm \
-e MEALIE_URL=http://your-mealie:9925 \
-e MEALIE_API_TOKEN=your-token \
mealie-mcp-server
# Run (HTTP with OAuth)
docker run --rm -p 3000:3000 \
-e TRANSPORT=http \
-e MEALIE_URL=http://your-mealie:9925 \
-e MEALIE_API_TOKEN=your-token \
-e MCP_SERVER_URL=https://mcp.your-domain.com \
mealie-mcp-server
Architecture
src/
├── index.ts # stdio entry point
├── http.ts # HTTP entry point (Streamable HTTP transport + OAuth)
├── server.ts # Creates MCP server and registers all tools (transport-agnostic)
├── mealie-client.ts # HTTP client for the Mealie REST API
├── types.ts # TypeScript type definitions for Mealie API shapes
├── auth/
│ ├── provider.ts # OAuth 2.0 provider (validates Mealie credentials, issues tokens)
│ └── login-page.ts # HTML login page rendered during OAuth authorization flow
└── tools/
├── get-recipe.ts # get_recipe tool
├── search-recipes.ts # search_recipes tool
├── create-recipe.ts # create_recipe tool
├── import-recipe.ts # import_recipe tool
└── update-recipe.ts # update_recipe tool
The server core (server.ts) is transport-agnostic — it builds the MCP server and registers tools without knowing which transport will be used. The two entry points (index.ts for stdio, http.ts for HTTP) each wire in their respective transport.
- stdio communicates via stdin/stdout using JSON-RPC 2.0. Standard for Claude Desktop and local MCP clients.
- Streamable HTTP exposes a
/mcpendpoint supporting POST (requests), GET (SSE stream), and DELETE (session termination). Uses stateful sessions withMcp-Session-Idheaders. Full OAuth 2.0 + PKCE authentication using Mealie credentials as the identity provider.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.