lunchmoney

This Lunchmoney MCP Server exposes four tools that let Claude and other AI assistants interact with your Lunchmoney data without exposing your full token or raw data. The tools include: get-recent-transactions (returns recent transactions within a specified number of days), search-transactions (find transactions by keyword in payee names or notes), get-category-spending (analyzes spending by category), and get-budget-summary (provides budget details including spending, remaining amounts, and recurring items). To use these tools, configure the MCP server locally and authorize access per request. Claude will query the server for specific information, and the server will fetch only the requested data from Lunchmoney and return concise results suitable for conversation.

When you ask Claude about your finances, you can prompt with natural-language questions like: "Show me my recent transactions from the past 7 days," "Find transactions at Amazon," "What did I spend in groceries this month?" or "Give me a summary of my monthly budget including remaining amounts." The server processes these requests locally, and only the relevant information is shared back with Claude, preserving privacy and token security on your device.

If you want more detailed analysis, you can combine tools in a single query, for example: "Show me my budget status for this month and list any categories that are over budget."

Prerequisites:

• Node.js (LTS) installed on your machine
• npm (comes with Node.js) or npm/yarn available
• A Lunchmoney API token from your Lunchmoney developer settings

Installation steps:

1. Install Node.js if you haven't already. Visit https://nodejs.org/ and install the LTS version for your OS.
2. Obtain your Lunchmoney API token from your Lunchmoney developer settings: https://my.lunchmoney.app/developers
3. Install the Lunchmoney MCP server package and run it via npx as described in the README:

# Ensure you have Node.js and npm installed
# Start the MCP server using npx (the package will be downloaded temporarily):
npx -y lunchmoney-mcp-server

4. Configure Claude Desktop (or your MCP client) to point to the server with the required environment variable:

• LUNCHMONEY_TOKEN: your Lunchmoney API token

Example configuration (as shown in the README):

{
  "mcpServers": {
    "lunchmoney": {
      "command": "npx",
      "args": ["-y", "lunchmoney-mcp-server"],
      "env": {
        "LUNCHMONEY_TOKEN": "your_token_here"
      }
    }
  }
}

5. Restart Claude Desktop or your MCP client if you updated the configuration.

Note: You can run the MCP server locally to ensure data never leaves your device; the server will fetch data from Lunchmoney on-demand and return only the requested results.

Tips and caveats:

• Keep your Lunchmoney token on your local machine; the MCP server never exposes your full data to Claude.
• If Claude reports a connectivity issue, verify that the token is valid and the local MCP server is running on the expected port.
• The server fetches data per request, so response times depend on Lunchmoney API latency.
• You can customize or rename the server in your Claude configuration by adjusting the mcpServers section; the tool names (get-recent-transactions, search-transactions, get-category-spending, get-budget-summary) correspond to the documented capabilities.
• If you see mismatched date ranges, ensure you follow the API notes: budget data should use month boundaries (e.g., 2024-01-01 to 2024-01-31).
• Environment variable security: treat LUNCHMONEY_TOKEN as a secret; avoid sharing this value in logs or screenshots.