← All articles

September 3, 2026 · Vedanshu Jain

How to Connect Claude, ChatGPT or Gemini to Your WooCommerce Store (MCP Guide)

Connect Claude, ChatGPT or Gemini to WooCommerce in fifteen minutes with MCP: enable a server, add it to your client, authorise a scoped user, start read-only.

You can connect Claude, ChatGPT or Gemini to a WooCommerce store in about fifteen minutes using the Model Context Protocol (MCP): enable an MCP server for the store, add it to your AI client’s configuration, authorise a scoped user, and start with read-only questions. This guide covers the two servers you are likely to use, WooCommerce core’s own MCP integration and Urumi’s operations server, what each lets the assistant do, and the safety habits that keep a live store safe. Last reviewed 3 September 2026.

What you need

• A WooCommerce store on 10.9 or later for the core MCP integration (developer preview since 10.3).
• An AI client that supports remote MCP servers: Claude (desktop and web), ChatGPT, Gemini, Cursor and most agent frameworks.
• A WordPress user created for the assistant, with only the capabilities it needs.

Option 1: WooCommerce core MCP

1. Enable the integration. In WooCommerce 10.9 the MCP integration is a developer-preview feature; enable it as described in the WooCommerce release notes for your version.
2. Create a scoped user. Add a WordPress user for the assistant (Shop Manager is a reasonable starting role; tighten it later) and generate an application password.
3. Add the server to your client. In Claude, ChatGPT or Gemini open the MCP or connectors settings, add a remote server with your store’s MCP endpoint, and supply the application password when prompted.
4. Test read-only first. Ask “how many orders did we take yesterday and what was the average order value?” and check the answer against WooCommerce Analytics before allowing any write.

With the core server the assistant can query products and orders, create and edit products, change order status and add order notes. It cannot see performance, logs, deploys or infrastructure.

Option 2: Urumi’s MCP server (stores on Urumi)

1. Enable MCP from your Urumi account (one config; the team can do it for you). The server exposes the same store context Urumi’s Revenue, Builder and Analytics AI use.
2. Add the Urumi server to your client’s MCP configuration alongside, or instead of, the core server.
3. Authorise from your Urumi account. Access is scoped to your store and can be revoked at any time.
4. Ask an operations question. “What regressed in checkout this week?” returns the offending change from APM traces. “Why is the ring builder slow on mobile?” returns the slow hook or query.
5. Request a fix. Changes arrive as a reviewed pull request through isolated staging with one-click rollback, the same path Builder AI uses. Nothing writes to production directly.

There is no per-token markup: you bring the Claude, ChatGPT or Gemini subscription you already pay for, and you can switch providers without changing the store.

Prompts that work well on day one

• “Which products went out of stock in the last 7 days and how many orders mentioned them?”
• “Draft product descriptions for the 12 SKUs with empty descriptions; do not publish, show me a table first.”
• “Summarise every order note added this week that mentions ‘refund’.”
• On Urumi: “Compare checkout p95 this week to last week and explain the difference.”
• On Urumi: “Show gross margin by product for the last 90 days and flag anything under 40%.”

Safety habits

• Scoped user, never an administrator account.
• Read-only for the first week; enable writes one ability at a time.
• Prefer servers that stage changes and offer rollback for anything beyond catalogue and order edits.
• Keep an audit trail: order notes and pull requests are your record of what the assistant changed.

Frequently asked questions

Does connecting Claude or ChatGPT to WooCommerce cost extra?

The core MCP integration is free with WooCommerce; you pay your AI provider for usage. Urumi’s server is included in the platform with no markup on model usage.

Can the assistant break my store?

Through the core server it can edit products and orders, so a scoped user and a read-only first week matter. Through Urumi, code and configuration changes go through staging, review and rollback, so a bad suggestion never reaches production.

Which client is best?

Any MCP-capable client works. Use the one your team already pays for; the store context is the same.

Background: WooCommerce MCP explained and the AI for WooCommerce overview.

Built by the people who built WooCommerce core.

We built WooCommerce core at Automattic — the parts that matter in production: performance, payments, reliability. Earlier, engineering at HackerRank (Y Combinator) through its enterprise scale-up. Naman led Payments and WooCommerce releases to 4.5M merchants; Vedanshu led HPOS, Taxes, and Shipping. Run by AI, overseen by the people who built WooCommerce core.

• Naman Malhotra - LinkedIn
• Vedanshu Jain - LinkedIn

Grow your store's revenue on Urumi.

The AI platform D2C brands use to grow revenue — built by the people who built WooCommerce core.

See the WooCommerce platform  ·  Start your WooCommerce store on Urumi  ·  Talk to the founders

Agent live · 99.99% uptime · shipping today.