Printful MCP server for your AI

The connect form asks for one field, Private token. Printful shows it on a single screen: Developer Portal, then Your tokens, then create a token. There is no OAuth popup, redirect URL or client secret involved, and no refresh cycle to keep alive. PopMCP holds the token server-side and sends it over HTTPS as an Authorization: Bearer header.

Both. The v2 side covers the catalog, orders, stores and webhooks, and those tool names carry a v2_ prefix so you can tell them apart at a glance. The v1 side brings sync products, e-commerce sync, the mockup generator, files, product templates and approval sheets. That split is the usual Printful stumbling block: catalog products and variants are v2 objects, while the products attached to your store are v1 sync products.

After the token is accepted you choose which stores to connect and which workspace each one belongs in. Store listings come back according to the token's access level, so a narrower token surfaces fewer stores than a broad one. Stores you leave out of that selection are not part of the connection.

Yes, when the connection is set to full access. It can create a draft order, update it, estimate its costs, confirm it for fulfillment and cancel it. Confirming is the step worth guarding, because it releases the draft into fulfillment, so keep approval-required-for-writes switched on. Read-only mode removes the write tools entirely and leaves lookups and reporting.

32 of the 106 operations load up front, which keeps the tool list in your client short enough to stay useful. The rest are searchable, so the agent pulls in something narrower, like a size guide or a print file lookup, only when a request calls for it. Flip the full-catalog toggle if you would rather have everything exposed at once.

Webhook triggers follow the order lifecycle: shipped, canceled, failed, refunded, put on hold, hold removed, needs design approval, and package returned. Product triggers fire when a store product finishes syncing, when its details or variants change, and when it is deleted, plus one for catalog stock going out of or back into availability. Two more run on change detection rather than webhooks: a new order arriving in the store, and Printful raising an approval sheet on suggested file changes.

No. PopMCP hosts it and gives you a Streamable HTTP MCP URL. There is nothing to deploy, no callback URL to register, and no infrastructure to keep running or patch.

Any client that supports a remote Streamable HTTP MCP server — Claude Desktop, Claude Web, ChatGPT custom connectors, Codex and the Codex CLI, OpenCode and Z.ai's ZCode, as well as editors such as Cursor, Windsurf and Zed. Your client signs in to PopMCP with OAuth; the provider credential never leaves the server.

Loading 106 tool definitions into every conversation burns context before the work starts. PopMCP loads a curated 32-tool set covering the everyday workflows and keeps the rest searchable, so your AI pulls a specialist tool only when a task needs it. You can switch a connection to the full catalog at any time.

A tool is something your AI calls — there are 106 here. A trigger is something PopMCP watches for you: 17 Printful events that can start work on the PopMCP side. Triggers never enter your AI client's context, and none of them ask you to paste a callback URL into Printful.