Paddle

Paddle MCP server

Give any AI assistant direct access to the Paddle API over the Model Context Protocol.

This server gives AI assistants like Claude, Cursor, and Windsurf direct access to Paddle API operations: products, prices, subscriptions, customers, transactions, and more. Paddle hosts it, so there's nothing to install or run. This repository holds the setup instructions only, with no source code.

Three tools are exposed to your assistant: search the API reference, execute chained API calls in a sandbox, and report_missing_tool for telemetry. A fourth tool, show_metrics_chart, draws a metric as an interactive chart in clients that support MCP Apps.

Endpoint

A single MCP endpoint. Each host serves one environment:

The examples below add both, as paddle-live and paddle-sandbox, so you can tell your assistant which one to use.

Authentication

Live

There are two ways to authenticate: sign in with OAuth, or pass a Paddle API key as a bearer token.

OAuth (recommended). Point your client at the endpoint with no credentials. The first time it connects, a browser window opens and you sign in to Paddle, so there's no API key to create, copy, or keep in a config file.

• A new connection gets read access by default, limited to what your Paddle user's role already permits.
• Change what a connection can do, or revoke it, from Paddle > Connectors > MCP in your dashboard.
• You may be asked to reauthorize from time to time.

API key. Prefer a long-lived credential, or running somewhere a browser can't open? Pass a Paddle API key in the Authorization header on every request:

Authorization: Bearer <your-paddle-api-key>

Scope the key to the permissions you actually need. They are the only limit on what the assistant can do.

Sandbox

Sandbox connects with a sandbox API key (prefixed pdl_sdbx_), passed in the Authorization header on every request. The sandbox server doesn't support OAuth.

Authorization: Bearer <your-sandbox-api-key>

Scope the key to the permissions you actually need. They are the only limit on what the assistant can do.

Connect your client

Drop one of these configurations into your AI client. paddle-live carries no credentials, so your client runs the OAuth flow on first connect. Replace SANDBOX_API_KEY with your sandbox key.

To connect to live with an API key instead, give paddle-live the same Authorization header as paddle-sandbox, with your live key.

Claude Code

claude mcp add --transport http paddle-live https://mcp.paddle.com/mcpclaude mcp add --transport http paddle-sandbox https://sandbox-mcp.paddle.com/mcp --header "Authorization: Bearer SANDBOX_API_KEY"

Then run /mcp inside Claude Code to start the browser sign-in.

Claude Desktop

In claude_desktop_config.json:

{  "mcpServers": {    "paddle-live": {      "command": "npx",      "args": ["-y", "mcp-remote", "https://mcp.paddle.com/mcp"]    },    "paddle-sandbox": {      "command": "npx",      "args": [        "-y",        "mcp-remote",        "https://sandbox-mcp.paddle.com/mcp",        "--header",        "Authorization: Bearer SANDBOX_API_KEY"      ]    }  }}

Codex

codex mcp add paddle-live --url https://mcp.paddle.com/mcpcodex mcp login paddle-live
export PADDLE_SANDBOX_API_KEY="SANDBOX_API_KEY"codex mcp add paddle-sandbox --url https://sandbox-mcp.paddle.com/mcp --bearer-token-env-var PADDLE_SANDBOX_API_KEY

Cursor

In ~/.cursor/mcp.json:

{  "mcpServers": {    "paddle-live": {      "type": "http",      "url": "https://mcp.paddle.com/mcp"    },    "paddle-sandbox": {      "type": "http",      "url": "https://sandbox-mcp.paddle.com/mcp",      "headers": {        "Authorization": "Bearer SANDBOX_API_KEY"      }    }  }}

VS Code

Press Cmd/Ctrl + Shift + P, choose MCP: Open workspace folder MCP configuration, then edit mcp.json:

{  "servers": {    "paddle-live": {      "type": "http",      "url": "https://mcp.paddle.com/mcp"    },    "paddle-sandbox": {      "type": "http",      "url": "https://sandbox-mcp.paddle.com/mcp",      "headers": {        "Authorization": "Bearer SANDBOX_API_KEY"      }    }  },  "inputs": []}

ChatGPT

ChatGPT connects via Plugins rather than a config file. Custom MCP apps need a paid plan (Plus, Pro, Business, Enterprise, or Edu) with Developer mode enabled. They aren't available on the free tier.

1. Turn on Developer mode under Settings > Security and login.
2. Open Plugins, then create one and set the MCP server URL to https://mcp.paddle.com/mcp or https://sandbox-mcp.paddle.com/mcp.
3. For live, select OAuth and sign in with Paddle when prompted. For sandbox, or to use a live API key, set Authentication to Custom header and paste Authorization: Bearer API_KEY.
4. In a new chat, choose Developer mode from the Plus menu and select the app.

For more detailed setup instructions, see the full installation guide.

Use it

Ask your assistant in plain language, and name the server so it picks the right environment. For example:

Using paddle-sandbox, set up a Pro plan at $79/month with a 14-day trial, then show me the products and prices.

Using paddle-live, investigate the failed payment for [email protected] and tell me whether it's recurring.

Using paddle-live, chart my revenue for the last 3 months by week.

For more example prompts and best practices, see the Paddle MCP server docs.