The Stripe MCP server lets AI agents interact with the Stripe API and search Stripe’s knowledge base, including documentation and support articles. It is hosted at https://mcp.stripe.com, and Stripe’s recommended setup is its agent plugin: install the Stripe CLI, run stripe agent setup, and it detects the agents you use and configures the server and Stripe’s skills for you. You can also add it by hand to Cursor, Claude, Claude Code, Codex, VS Code or any MCP client.
The tool list is broad. Rather than one tool per endpoint, it exposes API search and detail tools plus generic read and write tools that reach many Stripe API methods, along with account info, an analytics tool (in preview, and needing Sigma for natural-language SQL), documentation search, and an implementation planner that walks you through accepting payments or setting up billing. Authentication is OAuth when a client acts as you, or a restricted API key when an application needs its own persistent access. With OAuth you can grant different permissions to live mode and to sandboxes.
Because this touches money, the safety design matters. Stripe requires human confirmation before certain writes, such as refunds and outbound payments: you approve through a link, and unapproved requests expire after 24 hours. Administrators can enable or disable MCP access for the whole team, separately for live mode and sandbox, and revoke OAuth sessions. Stripe also warns to keep human confirmation on and to be careful combining it with other servers, because of prompt injection.
Verdict: very useful for building and debugging integrations. Our advice is to start in a sandbox and use restricted keys with only the permissions the agent needs. See the best MCP servers for developers and the MCP guide for background.
Sources: Stripe’s MCP documentation, checked on September 21, 2026.
