# Shopify Adds Browser-Agent Tools for Checkout

Published 2026-09-29 · AI.info Team · AI.info

> Shopify’s checkout WebMCP tools let browser agents read and update an active checkout, then submit an order after the buyer confirms it. The tools run in the buyer’s browser session, and they require no merchant configuration or separate API.

## Checkout moves into the browser agent’s reach

On September 28, 2026, Shopify added WebMCP tools that let browser-based agents read and update an active checkout in a buyer’s browser session. Agents can now assist beyond product discovery and cart management, continuing through checkout and order confirmation, the company said in its [developer changelog](https://shopify.dev/changelog/posts/webmcp-support-for-checkout).

The tools expose four actions. `navigate_to_storefront` takes the agent to a store; `get_checkout` reads checkout state, messages and post-completion order details; `update_checkout` changes supported fields; and `complete_checkout` submits the purchase after the buyer confirms. The tools run inside checkout-web and use the same state as the checkout interface.

## Buyer approval still gates the order

The new tools do not remove the shopper from the transaction. `complete_checkout` submits an order after buyer confirmation, and the agent hands control back whenever the flow requires the buyer’s input. Shopify’s examples include 3D Secure authentication and blocking user-interface extensions.

That handoff is a boundary between agent assistance and actions that still require a person. The agent can handle supported checkout updates, but the company’s description does not present checkout as a fully autonomous purchase process: buyer confirmation remains part of submission, and additional authentication or extension prompts can interrupt the agent’s work.

Shopify’s checkout documentation says the buyer sees the same checkout state as the agent and handles page interactions, including Shop Pay login or payment challenges. Before the agent calls `complete_checkout`, the buyer must confirm the order. If checkout presents a review step, the buyer reviews and authorizes submission on the page before the agent calls the tool again.

## No separate merchant integration

Merchants do not need to configure checkout or expose a new API for these browser tools. They operate against the active checkout session rather than creating a separate checkout flow. That keeps the agent’s actions tied to the checkout already presented to the shopper.

The checkout tools register on eligible checkouts. `get_checkout` can read the current checkout or an order receipt on the Thank you page. `update_checkout` can change buyer contact details, fulfillment choices, discount codes, declared fields and payment information; it does not place an order. Shopify’s documentation cautions that updates use PUT semantics, so an agent should build its requested state from a fresh checkout reading.

The scope differs from Shopify’s server-side agent tools. The company’s [WebMCP documentation](https://shopify.dev/docs/api/web-mcp) describes tools used by agents shoppers bring to a store in their browser; agents connecting directly to Shopify’s MCP servers follow a separate path. Shopify’s storefront WebMCP tools are available on Liquid storefronts and Hydrogen developer previews, while agent support for WebMCP is currently limited to Chromium-based browsers.

## From browsing to order confirmation

Before the checkout update, Shopify’s browser tools already covered storefront and cart actions. Adding checkout tools extends that browser-based sequence to supported checkout updates and order submission, without requiring merchants to install or configure another integration. The new step also gives agents access to order details after completion through `get_checkout`.

The practical reach depends on the browser agent supporting WebMCP and on the checkout not requiring a human action the agent cannot perform. When authentication or a blocking extension intervenes, Shopify’s specified behavior is explicit: control returns to the buyer. That limit is part of the feature’s design: the agent acts on the open checkout, but the buyer handles required page interactions and confirms the order.

## Sources

1. Original source · Release notes [Shopify Developer Changelog](https://shopify.dev/changelog/posts/webmcp-support-for-checkout) shopify.dev

Other pages read

2. [Shopify Developer Documentation · Checkout webmcp](https://shopify.dev/docs/agents/carts-and-checkout/checkout-webmcp) shopify.dev
3. [Shopify Developer Documentation · page 2](https://shopify.dev/docs/api/web-mcp) shopify.dev

## Related stories

- [Stripe opens Checkout to AI agents via WebMCP browser tools](https://ai.info/news/stripe-adds-webmcp-to-checkout-for-ai-agent-purchases)
- [Shopify Gives AI Agents Access to More Than 1 Billion Products](https://ai.info/news/shopify-shop-mcp-ai-agents-billion-products)
- [Sierra releases Poppy 1.0, with Shopify and Meta support](https://ai.info/news/sierra-personal-agent-protocol-poppy-1-0)
- [Walmart Opens Its Shopping Systems to AI Agents](https://ai.info/news/walmart-opens-shopping-systems-to-ai-agents)
- [Manus 2.0 Adds Event-Triggered Workflows and Personal Agents](https://ai.info/news/manus-2-adds-event-triggered-workflows-personal-agents)

## Explore

[More articles](https://ai.info/news)

---

Canonical: https://ai.info/news/shopify-webmcp-browser-agent-checkout
