Stripe
Read and write with the Stripe API methods Stripe's MCP server supports, with Stripe confirming the sensitive ones.
A Stripe connection gives a playbook's functions Stripe's MCP server. You sign in to Stripe and choose the accounts and sandboxes the connection may reach, and it acts as you, the Stripe user who connected, for those accounts and for no connected account. The server supports money movement, and Stripe guards it: certain writes, such as refunds and outbound payments, Stripe asks that user to confirm, and a write they approve is asked again the next time.
Available on Auto Cloud and on self-hosted runtimes.
What a playbook can do with Stripe
- Read Stripe data with the methods the server supports; Stripe's documentation lists them.
- Create and change Stripe data where the server offers a method for it.
- Move money where the server offers it, with Stripe's confirmation from the person who connected.
Start a Stripe playbook in a sandbox: the connection reaches the sandboxes you choose at sign-in, and a sandbox behaves as the live account does.
Stripe's MCP server lists its own tools when a function runs, and once you connect, your agent lists them with list_connection_tools.
Stripe's tools can pay, refund, transfer or pay out money.
An example workflow
A daily reconciliation. A schedule trigger runs the workflow at 07:00. A lookup function reads yesterday's charges, refunds and disputes through the server's read tools. A computation function totals them by product and currency and compares the totals with the figures in the spreadsheet the finance team keeps, read through a Google Sheets or an Excel connection. A reasoning function reads the differences and the notes in its prompt about the ones that are expected, foreign exchange and timing among them, and writes what needs a look. An interaction function posts the summary to the finance channel through a Slack connection. No refund is issued by this workflow; where one is due, a separate interaction function asks the finance lead first, and Stripe itself asks the person who connected to confirm the refund before it goes through. The run's history in Studio keeps every figure and every answer.
How to connect Stripe
- Sign in to Stripe and allow access to the accounts and sandboxes you choose. The consent screen shows the permissions listed below.
Then:
- Ask your agent to connect Stripe to Auto, or open Settings, Agents, Connections in Auto Studio and choose Stripe.
- On the Studio page, check the service and its access, choose which playbooks may use the connection, and approve it.
- Have your agent check one small read through the connection before a function depends on it.
What it needs
- A Stripe user in the account, whose permissions the connection has, and who will be asked by Stripe to confirm sensitive writes.
- Your agent connected to Auto and signed in as you, with permission to change your organization, if your agent adds the connection.
- On a self-hosted runtime, nothing to register: Stripe's authorization server lets Auto register a public client itself, or takes one you registered.
Permissions Auto asks for
- Read and change Stripe data in the accounts and sandboxes chosen, with the permissions granted at sign-in.
Good to know
Give a Stripe connection to the one playbook that needs it, under Only select playbooks, rather than to all. A tool that writes is marked as one that acts, so a lookup function cannot call it. What a function reads, a customer's details or an invoice, is kept in the run's history, where people with access to the playbook can read it; add the connection with content recording off if that history should hold no Stripe data.