Setup

Three steps. No server to host. No account to create.

1

Install uv

uv is a fast Python package runner. It handles everything automatically — no Python install needed.

macOS / Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Already have uv? Skip this step.

2

Get your YNAB API key

Go to YNAB Developer Settings and create a new personal access token.

Your API key stays on your machine — it's never sent anywhere except directly to YNAB's API.

3

Configure your AI platform

Python / uv
{
  "mcpServers": {
    "ynab": {
      "command": "uv",
      "args": ["tool", "run", "mcp-ynab"],
      "env": {
        "YNAB_API_KEY": "your-api-key-here"
      }
    }
  }
}
Config file locations

Claude Desktop (JSON)

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %LOCALAPPDATA%\Packages\Claude_pzs8sxrjxfjjc\LocalCache\Roaming\Claude\claude_desktop_config.json

Note: The "Edit Config" button in Claude opens the wrong path on Windows (known bug)

Claude Code (CLI)

User scope: ~/.claude.json

Use claude mcp add-json --scope user for global access

ChatGPT Desktop (JSON)

macOS: ~/Library/Application Support/ChatGPT/chatgpt_config.json

Windows: %APPDATA%\ChatGPT\chatgpt_config.json

Gemini CLI (JSON)

All platforms: ~/.gemini/settings.json

Having trouble? Check the troubleshooting docs.