home / mcp / paprika mcp server

Paprika MCP Server

Provides search, read, and update capabilities for Paprika recipes via MCP.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "briantkatch-paprika-mcp": {
      "command": "/Users/yourusername/Developer/paprika-mcp/.venv/bin/paprika-mcp",
      "args": [],
      "env": {
        "PAPRIKA_EMAIL": "[email protected]",
        "PAPRIKA_PASSWORD": "yourpassword",
        "PAPRIKA_USER_AGENT": "Paprika-MCP/1.0"
      }
    }
  }
}

You can run a Paprika MCP Server to let AI assistants search, read, and safely update your Paprika recipes. It provides fast, contextual access to your data while keeping credentials and interactions separated from your Paprika CLI.

How to use

You connect your MCP client to the Paprika MCP Server to perform three core actions: search recipes across titles, ingredients, categories, directions, and notes; read full recipe data including all metadata; and update recipes with find/replace using your confirmation. Start the local server, then point your MCP client to the local or remote endpoint. Use the server to supply recipe data to your AI workflows, while keeping credentials managed by the server.

How to install

Prerequisites: Python 3.10 or higher (Python 3.13 recommended) and a Paprika account with recipes. Node.js is optional but required for pre-commit hooks.

1) Create and activate a Python virtual environment for the server.

2) Install and run the setup to install the server and configure credentials.

3) If you prefer manual setup, install dependencies and configure credentials as described below.

Configuration and credentials

The server supports credentials via environment variables and a configuration file. Use the following options to authenticate with Paprika.

Option 1: Interactive setup to configure credentials.

Option 2: Create a manual config file at ~/.paprika-mcp/config.json with your email and password.

Option 3: Use environment variables PAPRIKA_EMAIL and PAPRIKA_PASSWORD.

As an MCP Server

To run the Paprika MCP Server locally and expose it to your MCP client, configure the server in your MCP client to start the local process.

{
  "mcpServers": {
    "paprika": {
      "command": "/Users/yourusername/Developer/paprika-mcp/.venv/bin/paprika-mcp",
      "args": []
    }
  }
}

User-Agent and client compatibility

If Paprika for Mac is installed, a compatible User-Agent string is created automatically. If not, set PAPRIKA_USER_AGENT or include a user_agent field in the config to ensure the MCP client and Paprika server communicate correctly.

Available tools (functions)

The server exposes a set of tools you can use through the MCP interface. These tools perform focused actions on your recipe data.

Security notes

Credentials are stored in the server's credential store and are used to authenticate requests to Paprika. Use environment variables or a local config file with secure permissions.

Troubleshooting and tips

If you run into issues with startup or authentication, restart your MCP client to clear any stdio process caches. Rebuilds are only necessary if you change the server code or its entry points.

Available tools

format_fraction

Format a fraction string to unicode fraction characters, with local-only operation and optional regex support.

search_recipes

Search recipes by text across multiple fields including name, ingredients, categories, directions, and notes, with optional context lines.

read_recipe

Read full recipe data by ID or title, with Unicode normalization for title matching.

update_recipe

Update a specific field in a recipe using a find/replace approach, with an option to use regex for the find pattern.