@opencommerceprotocol/adapter-shopify

Shopify Storefront API adapter for the Open Commerce Protocol. Maps OCP tool calls to Shopify's GraphQL Storefront API, making your Shopify store discoverable and shoppable by AI agents.

Installation

npm install @opencommerceprotocol/adapter-shopify

Quick Start

import { OCP } from '@opencommerceprotocol/runtime';
import { createShopifyHandlers } from '@opencommerceprotocol/adapter-shopify';

OCP.init({
  handlers: createShopifyHandlers({
    storeUrl: 'https://mystore.myshopify.com',
    storefrontAccessToken: 'your-storefront-access-token',
  }),
});

Configuration

createShopifyHandlers accepts a ShopifyConfig object:

| Option | Type | Required | Description | |--------|------|----------|-------------| | storeUrl | string | Yes | Your Shopify store URL (e.g. https://mystore.myshopify.com) | | storefrontAccessToken | string | Yes | Shopify Storefront API access token | | apiVersion | string | No | Storefront API version (defaults to 2024-01) |

You can generate a Storefront access token in the Shopify Admin under Settings > Apps and sales channels > Develop apps.

API Reference

The adapter implements the following OCP tool handlers:

| Handler | Description | |---------|-------------| | search_products | Search products via Shopify's GraphQL products query. Supports query, category, in_stock, and limit params. | | get_product | Fetch full product details including images, variants, and ocp.agent_notes metafield. | | get_product_qa | Answer questions about a product using its description and agent notes. | | compare_products | Compare multiple products by fetching each and returning them together. | | add_to_cart | Create a Shopify cart via the cartCreate mutation. | | get_cart | Retrieve current cart contents. | | update_cart | Update cart item quantities. | | begin_checkout | Return a checkout URL for the store. | | check_availability | Check whether a product is in stock. |

Agent Notes via Metafields

The adapter reads the Shopify metafield namespace: "ocp", key: "agent_notes" on each product. Use this to provide AI-optimized product descriptions that help agents make better recommendations.

License

Apache-2.0