Checkout Page MCP

The Checkout Page MCP server lets AI assistants like Claude, ChatGPT, and Cursor work directly with your Checkout Page account. Instead of digging through the dashboard or writing API requests, you can just ask, in plain English, and the assistant takes care of the rest.

It's built on the Model Context Protocol, an open standard for connecting AI assistants to the tools and data they need.

The server URL is:

What can the Checkout Page MCP do?

Once connected, you can ask your AI assistant to do things like:

• "Find the customer with the email jane@example.com and show me their last subscription."
• "Pull up the most recent payment for order ABC123. What plan did they pick?"
• "Show me the last 10 form submissions from my ebook lead capture form."
• "Open the checkout page with ID cp_abc123 and tell me what product is attached."
• "Spin up a new checkout page called 'Spring Workshop' priced at $99."
• "Create a lead capture form for our new ebook with name, email, and company fields."
• "Build a registration form for the May webinar. Collect name, email, and which session they want to attend."
• "Create a 20% off coupon called SPRING20 that expires May 31."
• "Add the Spring Workshop checkout to the pricing section of the site we're building, as a popup button."

Your assistant uses the tools below to answer these questions and take action on your account.

Available tools

Before you start

You'll need a Checkout Page account. The MCP server uses OAuth, so there are no API keys to copy or paste. Your AI tool walks you through a one-click sign-in the first time you connect.

Connect your AI tool

Each AI tool has its own settings screen for adding a connector. Pick yours below. The steps take about a minute. Once connected, try the test prompt at the end of each section to confirm it's working.

ChatGPT

1. Open ChatGPT.com and go to Settings → Security and login. Enable Developer mode.
2. Close settings and go to chatgpt.com/plugins.
3. Click the + in the top right. (Custom plugins are available on paid plans.)
4. Enter:
   • Name: Checkout Page
   • MCP server URL: https://mcp.checkoutpage.com
   • Authentication: OAuth
5. Toggle I understand and want to continue on.
6. Click Create, then follow the Checkout Page OAuth flow to connect.
7. Start a new chat.
8. Click + and search for Checkout Page.

Test it: ask "List my last 5 payments from Checkout Page."

Claude

1. Open Claude and go to Settings → Connectors.
2. Click Add custom connector.
3. Enter:
   • Name: Checkout Page
   • MCP server URL: https://mcp.checkoutpage.com
4. Click Add, then click Connect on the Checkout Page connector and sign in.

Test it: ask "Show me the last 5 form submissions from Checkout Page."

Cursor

The fastest way is the one-click install:

Or add it manually:

1. Open Cursor, then go to Settings → MCP.
2. Click Add new MCP server.
3. Enter:
   • Name: checkoutpage
   • URL: https://mcp.checkoutpage.com
4. Click Save, then sign in via the OAuth prompt.

Test it: in Cursor's chat, ask "List the last 5 customers from Checkout Page."

VS Code

GitHub Copilot Chat and the official MCP extension both support our server.

1. Open VS Code and run the MCP: Add Server command from the Command Palette (Cmd/Ctrl + Shift + P).
2. Choose HTTP.
3. Enter the URL: https://mcp.checkoutpage.com
4. Name it: checkoutpage
5. Sign in via the OAuth prompt that opens in your browser.

Test it: in Copilot Chat, ask "Show me my Checkout Page subscriptions."

Codex

1. Open Codex and go to Settings → MCP servers.
2. Click + Add server.
3. Choose Streamable HTTP as the transport.
4. Enter:
   • Name: Checkout Page
   • URL: https://mcp.checkoutpage.com
5. Click Save, then sign in via the OAuth prompt.

Test it: ask "List my last 5 payments from Checkout Page."

Claude Code

Two steps: add the server, then sign in from inside Claude Code.

claude mcp add --transport http checkoutpage https://mcp.checkoutpage.com

Then start a session:

claude

In the session:

1. Type /mcp.
2. Use arrow keys to highlight checkoutpage and press enter.
3. Choose authenticate and follow the OAuth prompt.

Test it: ask "List my last 5 payments from Checkout Page."

Other AI tools

Any client with MCP support can connect to https://mcp.checkoutpage.com using OAuth. If your tool asks for a transport, choose Streamable HTTP. Most clients accept this config:

{
  "mcpServers": {
    "checkoutpage": {
      "url": "https://mcp.checkoutpage.com"
    }
  }
}

Example prompts

Launch a new product

You're prepping for a new workshop and want a checkout page and early-bird discount live in minutes.

Paste this into your AI tool:

Create a checkout page called "Pricing Strategy Workshop" priced at $199. Write a confident product description (3–4 sentences) that pitches a half-day workshop for founders and product leaders who want to raise prices without losing customers. Turn on the coupon code field so customers can apply discounts. Then create a coupon code EARLYBIRD that takes 20% off and expires May 15. Send me both the checkout URL and the coupon code when you're done.

The assistant calls create_checkout_page (with a written description and the coupon field enabled), then create_coupon, and returns the live checkout URL and the discount code.

You can use the same pattern for any new offer: a course launch, a beta signup page, or a limited-time promotion.

Add a checkout to a website you're building

You're building a site with an AI tool like Lovable, Bolt, v0, Cursor, or Claude Code and want people to buy right on the page.

Paste this into your AI tool:

Create a checkout page called "Pricing Strategy Workshop" priced at $199, then add it to the pricing section of this website as an inline embed.

The assistant calls create_checkout_page, then add_to_website, and places the embed code in your site. It uses a React component when the site is built with React, and plain HTML otherwise. Ask for a popup button instead if you'd rather keep customers on the page until they click.

Reply to a customer billing question

A customer emails asking when their next payment will hit and you want a draft reply you can send back in seconds.

Paste their question into your AI tool:

A customer named Jane Smith (jane@example.com) just emailed asking: "When is my next payment, and how much will I be charged?" Look up her account in Checkout Page, find her active subscription, and draft a friendly reply I can send back. Include the next billing date and the amount.

The assistant calls list_customers to find Jane by email, then list_subscriptions to pull her active plan, and drafts a ready-to-send reply with the next billing date and amount.

You can use the same pattern for refund requests, receipt re-sends, or any one-off support question that needs account context.

Test your checkout

Before you launch, you want to buy from your own checkout and see exactly what a customer gets, without paying.

Paste this into your AI tool:

Test my "Pricing Strategy Workshop" checkout page in Checkout Page test mode. Put the page in test mode, open it in the browser and buy it with the test card 4242 4242 4242 4242 and a new email address. Then find the test payment it created and check that it's marked livemode false. Switch test mode back off when you're done, and tell me what you bought and what you found.

The assistant calls list_checkout_pages to find the page and update_checkout_page with testmode: true, pays in its browser, calls list_payments with livemode: false to find the test payment, then calls update_checkout_page again to switch test mode off.

You can use the same pattern for an event or a form, or to check a webhook or integration fires before you rely on it.

Tips for better results

• Ask the AI to write copy for you: product titles, descriptions, confirmation messages. Otherwise pages come out blank or generic.
• Be explicit about settings the AI can't guess (e.g. "turn on the coupon code field").
• Stick to specific records and creation tasks. AI assistants are less reliable at totals and rankings ("top 5 pages by revenue") because that requires reading every payment. For numbers like that, use the dashboard's analytics view.
• Some checkout settings, like custom button text or page styling, aren't exposed via MCP yet. If a request doesn't take effect, open the page in your dashboard to finish the tweak.

Test mode

Your assistant can make test purchases on a checkout page, event or form without paying, then check what they created. See Test mode for how it works.

• Switch it on and off. create_checkout_page, create_event and create_form, and the matching update_* tools, take testmode: true or testmode: false. list_checkout_pages, list_events and list_forms take testmode: true to find pages still in test mode.
• Read test records. The list and get tools for payments, bookings, subscriptions, subscription payments, invoices and submissions, and list_tickets, take livemode: false. Without it they return live records only.

When your AI tool connects, the server also gives your assistant a short primer on test mode, so you can ask in plain words.

Test prompts

Two prompts script a full test run. They show up as slash commands or prompts in AI tools that support MCP prompts. In Claude Code, for example, type /mcp__checkoutpage__ and pick one.

For example, to test an event in Claude Code, type the prompt name, the event's id and its page type:

/mcp__checkoutpage__test_checkout {eventId} event

For funnels, see Test an upsell funnel.

To pay, your assistant needs a browser tool, such as Claude in Chrome, a Playwright MCP server or ChatGPT agent mode. Without one, it gives you the link to pay yourself, then carries on checking the records.

Troubleshooting

Security and privacy

• The MCP server only acts on your behalf after you sign in via OAuth.
• Tokens are scoped to your account. Revoke access anytime from your Checkout Page settings.
• Read tools (list_*, get_*) never modify data. Write tools (create_*, upload_*) and destructive tools (delete_*) are clearly named so your assistant can ask before making changes. Deleting a webhook is irreversible, so confirm before your assistant runs delete_webhook.
• Data your assistant looks up (customer details, payment info) is sent to your AI tool to generate a response, the same way it works for any other connector. Choose AI tools whose data-handling policies match your needs.