Stripe MCP Tools for AI Agents
Let an agent create customers, invoices, products, payment links, and refunds, or read Stripe balances and payment data.
Available Tools
15 tools available in this integration
Stripe.CreateBillingPortalSession This tool will create a billing portal session.
Stripe.CreateCustomer This tool will create a customer in Stripe.
Stripe.CreateInvoice This tool will create an invoice in Stripe.
Stripe.CreateInvoiceItem This tool will create an invoice item in Stripe.
Stripe.CreatePaymentLink This tool will create a payment link in Stripe.
Stripe.CreatePrice This tool will create a price in Stripe.
Stripe.CreateProduct This tool will create a product in Stripe.
Stripe.CreateRefund This tool will refund a payment intent in Stripe.
Stripe.FinalizeInvoice This tool will finalize an invoice in Stripe.
Stripe.ListCustomers This tool will fetch a list of Customers from Stripe.
Stripe.ListInvoices This tool will list invoices in Stripe.
Stripe.ListPaymentIntents This tool will list payment intents in Stripe.
Stripe.ListPrices This tool will fetch a list of Prices from Stripe.
Stripe.ListProducts This tool will fetch a list of Products from Stripe.
Stripe.RetrieveBalance This tool will retrieve the balance from Stripe.
Working directly against the raw API? 220 more tools in the Stripe API server →
Why agent-optimized tools matter
Built for intent, not API schemas
Ask Stripe to "find the overdue invoice for Acme and prepare it for payment." Arcade helps the agent use the right Stripe actions without guessing customer, invoice, or price identifiers.
Up to 100x fewer tokens
Stripe list tools let the agent inspect customers, invoices, payment intents, products, and prices without loading unrelated collections into context. It gets the records needed for the next action instead of the whole account.
Complete the workflow end to end
Find the customer, inspect the invoice, add a line item, finalize the invoice, and create a payment link through one toolkit designed around the complete billing task.
Make actions predictable and controlled
Arcade tools return structured results and useful errors across users and runs. Teams can apply scope restrictions, redaction, and approval policies so agents follow the same controls every time.
Securing Stripe
The more capabilities agents get, the more useful they are. However, more capabilities bring more risks of undesired effects such as data loss and exfiltration.
This is why Arcade provides you with contextual rules that dictate how and when a tool is allowed to run. Arcade’s contextual access lets you govern tool visibility and behavior at every stage of execution and control who can see a tool, what inputs are allowed, and what comes back.
Learn about Contextual Access PoliciesFor Stripe, you can implement some of these policies:
- Hide Stripe.CreateRefund from agents unless the roles your IDP marks as billing owners are active, so an agent cannot return customer payments without authorization.
- Restrict Stripe.CreatePaymentLink to agents operating under your sales-operations role, so unreviewed automation cannot publish a new way to charge customers.
- Deny Stripe.FinalizeInvoice to agents outside your accounts-receivable role, so an agent cannot turn an unreviewed draft into a customer-facing invoice.
- Require human approval before Stripe.CreateRefund runs, so a mistaken or fraudulent refund cannot move money without review.
- Require human approval before Stripe.CreatePaymentLink runs, so an agent cannot publish a payment page with an unapproved price.
- Block Stripe.FinalizeInvoice when the invoice lacks an approved billing owner, so an agent cannot send an incorrect amount to a customer.
- Redact email addresses, phone numbers, and billing addresses from Stripe.ListCustomers results before they reach the model.
- Strip invoice descriptions, customer contact details, and payment terms from Stripe.ListInvoices results, so the model receives invoice status without unnecessary customer data.
- Deny the output of Stripe.RetrieveBalance to agents outside your finance role, so account balances do not reach unauthorized workflows.
How to use Stripe with your agent
Pick your MCP client below and follow the setup to connect your agent to Stripe. Then try prompts that create, read, and manage payment data.
Your questions, answered
A Stripe MCP server exposes Stripe capabilities as structured tools that an AI agent can call through the Model Context Protocol. With the required API key or secret, the agent can perform permitted actions such as listing payment intents, creating invoices, issuing refunds, and creating payment links.
Stripe's own MCP server is a first-party endpoint for accessing Stripe tools. Arcade adds the actions runtime around Stripe, including authenticated access, agent-optimized tools, approval workflows, redaction, scope controls, and access from supported frameworks and MCP clients.
Arcade's built-in Stripe tools can create customers, products, prices, invoice items, invoices, payment links, billing portal sessions, and refunds; finalize invoices; list customers, invoices, payment intents, prices, and products; and retrieve the Stripe balance. Arcade also provides access to 220 raw Stripe API tools.
Stripe permissions come from the shared API key or secret configured for the integration. Use a restricted Stripe key with only the customer, product, invoice, payment, balance, and refund access your agent needs; Arcade does not replace Stripe's key permissions.
No. Stripe uses one shared API key or secret configured once for the integration, rather than having each user connect a separate Stripe account. Arcade brokers the configured credential, and the model never receives it.
Yes. Contextual Access policies can require human approval before consequential Stripe actions such as creating a refund, finalizing an invoice, creating a payment link, or creating other records. The call proceeds only after the policy's conditions have been satisfied.
Arcade proxies Stripe API calls and does not persist your Stripe content as part of normal tool execution. The configured credential is brokered and stored encrypted at rest, and the model never receives the raw API key or secret.
Any framework or client that can connect to Arcade's MCP gateway or tool APIs works, including LangChain, OpenAI Agents, CrewAI, Claude Desktop, Claude Code, Cursor, Visual Studio Code, Microsoft Copilot Studio, and GitHub Copilot.
Arcade can use the configured Stripe credential while it remains valid. If the key is revoked, rotated, or otherwise stops working, Stripe calls stop until the credential is reconfigured.
Stripe's API rate limits and any account- or key-level quotas still apply to calls made through Arcade. Arcade does not remove Stripe's limits, so high-volume workflows must stay within the limits attached to the Stripe account and credential.