MCP Ecommerce

Store platform

BigCommerce MCP Server (Storefront)

BigCommerce's official storefront MCP server gives AI agents product search, product details, cart handling and checkout links in a guest shopping flow. It is in private beta.

BigCommerce MCP Server (Storefront) setup guide: BigCommerce's MCP documentation page on docs.bigcommerce.com, shown in a browser frame (store platform MCP server for e-commerce)
BigCommerce MCP Server (Storefront) setup guide preview: BigCommerce’s documentation on docs.bigcommerce.com, captured Oct 13, 2026. The page may have changed since; the link opens the live version.

What it is and why it matters

The BigCommerce Storefront MCP server is the vendor's official, hosted MCP endpoint for a store's front end. It lets an AI agent search the catalog, read product details, build and change a cart, and hand the shopper a checkout link. The shopper finishes payment on the store itself. Each storefront gets its own MCP server URL, which the store owner generates in the control panel. The listing marks it as private beta.

It is aimed at BigCommerce merchants and agencies who want to test AI shopping assistants, not at people who want reporting or store administration. The documentation describes a B2C tool set for guest and logged-in shoppers, plus a B2B tool set for authenticated Buyer Portal users, covering shopping lists and quotes, gated by B2B Edition and buyer permissions. Whether it works well depends on beta access, on a store owner enabling it, and on how complete and clean your catalog data is, because the agent can only describe what the storefront exposes.

What it does

  • Search the catalog and fetch product details
  • Create and update a cart for a guest shopper
  • Generate checkout links the shopper completes on the store
  • Serve as the agentic commerce entry point for a BigCommerce store

Best for

BigCommerce merchants and agencies preparing for AI shopping assistants, once they have beta access.

How to connect it

  1. Confirm you have beta access and a BigCommerce store with an active storefront. The listing says access is required, so check with BigCommerce if the MCP option does not appear.
  2. Have the store owner open the BigCommerce control panel, go to Settings, and under API and MCP select MCP. Only a store owner can enable it, as described in the documentation.
  3. Turn on the Storefront MCP toggle and accept the terms of service. The page then shows who accepted the terms and when.
  4. Under Select storefront, choose the storefront and copy its MCP API Endpoint. Each storefront has its own URL. Allow up to 10 minutes for the server to become fully available.
  5. In your AI client (Claude Desktop, Claude Code, Cursor or Warp), add the server as described in the documentation. For Claude Desktop, this means editing the config under Settings, Developer, and using the mcp-remote package with your URL. This needs Node.js 18 or higher.
  6. Test with a guest shopping flow on a non-critical storefront first. If you are building a chat widget on a Stencil storefront and want logged-in shoppers, follow the vendor's Storefront session sync documentation.

Exact commands and addresses change. Use the official documentation for the current steps.

Installation guide for BigCommerce MCP Server (Storefront)

Client-by-client steps, what to prepare, the first prompts to run and fixes for common errors.

ClaudeChatGPTGeminiMicrosoft CopilotPerplexityCursorClaude Code
→

Questions you can ask it

  • Search the catalog for waterproof hiking jackets and show me the options with prices.
  • Show me the details for the product called Trail Runner 2, including its available variants.
  • Create a cart with two of the medium blue version of this product and give me the checkout link.
  • Change the quantity of the first item in my cart to one and tell me what the cart contains now.

What it cannot do

  • It is a shopper-facing server, so it does not provide sales reports, order management or store administration.
  • It does not take payment inside the chat. It produces a checkout link that the shopper completes on the store.
  • It requires private beta access and the tool set is still limited, so some storefront features may not be available to the agent.

Staying safe

  • Only a store owner can enable Storefront MCP and accept the terms. Record who accepted them and treat the endpoint URL as something to share only with your development team or trusted partners.
  • The cart tools change cart data, so test on a staging or low-risk storefront and supervise early sessions. Because checkout stays on the store, the agent does not handle card details.
  • Session sync links the agent to a logged-in customer or B2B buyer, including company and permissions. Enable it only after reviewing what that customer data exposes, and test with dummy accounts first.

Watch out for: Private beta: access is required and the tool set is limited so far. It is a shopper-facing server, not a reporting or admin tool.

Frequently asked questions

+Is the BigCommerce MCP server available to every store?

Not yet. The listing describes it as a private beta, so you need access before it works. Once you have it, a store owner enables Storefront MCP under Settings, API and MCP, accepts the terms, and copies the storefront's endpoint URL. Check with BigCommerce about access.

+Can the BigCommerce MCP server read my orders or sales reports?

No. It is built for shopping: product search, product details, carts and checkout links. It is not a reporting or admin tool, so use other BigCommerce tools or the control panel for orders, analytics and store management.

+Which AI clients can connect to the BigCommerce MCP server?

The documentation lists Claude Desktop, Claude Code, Cursor and Warp. The connection uses the mcp-remote package through npx, which needs Node.js 18 or higher, and your storefront's MCP URL. Follow the vendor documentation for the exact setup for your client.

Step-by-step: install BigCommerce MCP Server (Storefront) in Claude, ChatGPT, Gemini, Copilot and more →

Related guides

Other servers to compare