Shopify MCP
A Model Context Protocol server that connects AI agents to the Shopify Admin GraphQL API, enabling management of products, orders, customers, and more through natural language.
README
███████╗██╗ ██╗ ██████╗ ██████╗ ██╗███████╗██╗ ██╗ ███╗ ███╗ ██████╗██████╗
██╔════╝██║ ██║██╔═══██╗██╔══██╗██║██╔════╝╚██╗ ██╔╝ ████╗ ████║██╔════╝██╔══██╗
███████╗███████║██║ ██║██████╔╝██║█████╗ ╚████╔╝ ██╔████╔██║██║ ██████╔╝
╚════██║██╔══██║██║ ██║██╔═══╝ ██║██╔══╝ ╚██╔╝ ██║╚██╔╝██║██║ ██╔═══╝
███████║██║ ██║╚██████╔╝██║ ██║██║ ██║ ██║ ╚═╝ ██║╚██████╗██║
╚══════╝╚═╝ ╚═╝ ╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═════╝╚═╝
Shopify MCP
A Model Context Protocol (MCP) server that connects agents to the Shopify Admin GraphQL API. Use it to browse, edit, and clean up store data via a curated set of tools.
npm: shopify-mcp
binary: shopify-mcp
Highlights
- CRUD for products, collections, orders, and customers
- Draft orders for quotes, manual orders, and B2B pricing
- Inventory and location lookups for stock workflows
- Metafields for custom data
- Metaobject entry creation and lookup for existing definitions
- URL redirects management
- OAuth login flow with local token caching
- Bulk product cleanup utilities
Prerequisites
- Node.js 18+
- A Shopify custom app (OAuth or Admin API token)
Local setup (this repo)
Use this when you want to run the MCP server from this local checkout instead of a remote deployment.
- Install dependencies and build:
npm install
npm run build
- Create local env config:
cp .env.example .env
Set at least:
MYSHOPIFY_DOMAIN=your-store.myshopify.comSHOPIFY_ACCESS_TOKEN=shpat_xxxREMOTE_MCP=false
- Start local MCP (stdio):
npm run start:local
start:local uses stdio mode. Remote mode is only enabled with --remote or REMOTE_MCP=true.
Install + run
OAuth flow (recommended)
- Create a custom app and copy Client ID and Client Secret.
- In App setup, set App URL and Allowed redirection URLs to:
http://localhost:3456/callback - Start the OAuth flow:
npx shopify-mcp --oauth --domain=your-store.myshopify.com --clientId=xxx --clientSecret=yyy
Tokens are stored at ~/.shopify-mcp/tokens.json. After that, start the server with just the domain:
npx shopify-mcp --domain=your-store.myshopify.com
Optional: override scopes with --scopes or SHOPIFY_SCOPES.
Access token (manual)
- Create a custom app in Shopify
- Enable Admin API scopes:
read_products,write_productsread_customers,write_customersread_orders,write_ordersread_draft_orders,write_draft_ordersread_inventory,write_inventoryread_locationsread_content,write_contentread_files,write_files
- Install the app and copy the Admin API access token
Run:
shopify-mcp --accessToken=<YOUR_ACCESS_TOKEN> --domain=<YOUR_SHOP>.myshopify.com
MCP client setup
Claude Desktop (local repo build)
Build first (npm run build), then point Claude Desktop at this repo's built entrypoint:
{
"mcpServers": {
"shopify-local": {
"command": "node",
"args": [
"/absolute/path/to/shopify-mcp/dist/index.js",
"--domain",
"your-store.myshopify.com"
],
"env": {
"SHOPIFY_ACCESS_TOKEN": "shpat_xxx",
"REMOTE_MCP": "false"
}
}
}
}
If you completed OAuth locally, remove SHOPIFY_ACCESS_TOKEN and keep --domain.
Claude Desktop (npm package)
{
"mcpServers": {
"shopify": {
"command": "npx",
"args": [
"shopify-mcp",
"--accessToken",
"<YOUR_ACCESS_TOKEN>",
"--domain",
"<YOUR_SHOP>.myshopify.com"
]
}
}
}
If you completed OAuth, omit --accessToken and keep --domain.
Config paths:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%/Claude/claude_desktop_config.json
Remote MCP (Railway, etc.)
By default this server runs as a local stdio MCP. Passing --remote (or setting
REMOTE_MCP=true) switches it to HTTP/SSE mode so it can be deployed as a remote
MCP server for Claude.ai or other remote clients. This repo ships a Dockerfile
and railway.json so Railway builds and starts it in remote mode out of the box.
1. Get a token locally (one-time):
npx shopify-mcp --oauth --domain=your-store.myshopify.com --clientId=xxx --clientSecret=yyy
# Token saved to ~/.shopify-mcp/tokens.json
2. Deploy to Railway:
- Create a project from this repo. Railway reads
railway.jsonand builds theDockerfile, which starts the server with--remote. - Set the service environment variables (Railway injects
PORTautomatically):
SHOPIFY_ACCESS_TOKEN=shpat_xxx # from tokens.json (or use the OAuth vars)
MYSHOPIFY_DOMAIN=your-store.myshopify.com
MCP_API_KEY=choose-a-long-random-string # required to authenticate remote clients
# REMOTE_MCP=true is already implied by the Dockerfile's --remote flag
# PORT is injected by Railway (defaults to 3000 when run locally)
3. Connect:
- Health check:
GET /health - MCP endpoint (SSE):
GET /mcp?apiKey=<MCP_API_KEY> - Messages:
POST /messages?apiKey=<MCP_API_KEY>
Auth uses the apiKey query parameter; requests without a matching
MCP_API_KEY receive 401.
Test the container locally before deploying:
docker build -t shopify-mcp .
docker run -p 3000:3000 \
-e MYSHOPIFY_DOMAIN=your-store.myshopify.com \
-e SHOPIFY_ACCESS_TOKEN=shpat_xxx \
-e MCP_API_KEY=test \
shopify-mcp
# then in another shell: curl localhost:3000/health
Environment variables (optional)
SHOPIFY_ACCESS_TOKEN=your_access_token
MYSHOPIFY_DOMAIN=your-store.myshopify.com
# Optional OAuth values:
# SHOPIFY_CLIENT_ID=your_client_id
# SHOPIFY_CLIENT_SECRET=your_client_secret
# SHOPIFY_SCOPES=comma,separated,scopes
Tool catalog
Products
products— unified lookup/search/filter. Passidfor a single product; omitidto list/search with filters (title,status,vendor,tag, inventory, dates,hasImages, …). Returns the product's Shopify Standard Product Taxonomycategory({id, name, fullName}) inslim/standard/full. Page size capped at 100.create-productupdate-product— acceptscategory(Shopify Standard Product Taxonomy GID,vp-*prefix); the tool verifies the category actually stuck and throws a loud, actionable error if Shopify silently rejected the GID, instead of leaving you with a nullcategory.delete-productdelete-variantdelete-product-imagesbulk-update-productsbulk-delete-productscount-products-by-tagfind-products-by-metafield— list products that have / don't have / both for a givennamespace.key, paginated across the whole catalog via cursorsearch-taxonomy— browse Shopify's product category taxonomy; setincludeAttributes:trueto also return each category's attributes (e.g. Color, Pattern) and their allowed values
Collections
get-collectionsmanage-collection-productscreate-collectionupdate-collectiondelete-collection
Customers
get-customers(supports pagination viacursor)update-customer
Orders
orders— unified lookup/list. Passidfor a single order; omitidto list with filters (customerId,status, pagination viacursor). Replacesget-orders,get-order-by-id, andget-customer-orders.update-order
Draft Orders
draft-orders— unified lookup/list. Passidfor a single draft order; omitidto list with filters (status,query, pagination viacursor).create-draft-orderupdate-draft-ordercomplete-draft-order
Inventory
get-inventory-levelsupdate-inventory
Locations
get-locations
Metafields
get-metafields— server-side filter withkey+namespace(single field) orkeys: ["namespace.key", …](multi) via Shopify's nativemetafields(keys:); setincludeDefinitions:trueto merge ALL definitions with current values so empty/unfilled fields show up (value:null,isSet:false)set-metafield(create or update; supportsmetaobject_reference/list.metaobject_reference)bulk-set-variant-metafields— set metafields across many variants of one product in a singleproductVariantsBulkUpdatecall (up to 250 variants/call). UNIFORM mode (metafields) fans one value out to every variant and auto-discovers the variant IDs; PER-VARIANT mode (variants) sets different values per variant. Avoids oneset-metafieldcall per variant.delete-metafieldlist-metafield-definitions— discover metafield definitions for an owner type (PRODUCT, ORDER, CUSTOMER, …); each entry now includesconstraints(e.g.{key:"category", values:["vp-2","vp-2-2-3", …]}) so agents can see category-gating before writing (e.g.vehicle_*requiresvp-2*Vehicle categories; values on disallowed categories are silently filtered out by Shopify on read).get-metafield-options— resolve a metafield's selectable options in one call (for metaobject-reference fields, returns the available metaobject entries; for choice-lists, the allowed choices)
Metaobjects
list-metaobject-definitionsget-metaobject-definitioncreate-metaobject— optionalstatus(ACTIVE/DRAFT); defaults to Shopify'sDRAFTfor publishable definitions, passACTIVEto publish on createupdate-metaobject— edit fields on an existing entry (only provided keys change); optionalstatusto publish (ACTIVE) or unpublish (DRAFT)delete-metaobjectlist-metaobjects— returnsstatusper entry; optionalstatusfilter (applied client-side to the fetched page)get-metaobject— returns the entry's publishstatus
Files
get-files— list/search files in the storeattach-file-to-product— attach an existing media file to a productdetach-file-from-product— remove a media file from a productcreate-file-upload-session— start a browser upload session (remote mode only)get-file-upload-session— check an upload session (remote mode only)
URL redirects
get-redirectscreate-redirectdelete-redirect
Analytics
get-store-counts- Get all key counts in one call (products, variants, orders, customers, collections)get-product-issues- Audit products for problems (zero inventory, low stock, missing images, zero price)
Bulk Operations
start-bulk-export- Start async bulk export (products, orders, customers, inventory, or custom query)get-bulk-operation-status- Check progress of bulk operationget-bulk-operation-results- Download and parse completed results (summary, sample, or full)
Server
get-status- Report MCP server status, configured store, and connection health
Debugging
Tail Claude Desktop logs:
tail -n 20 -f ~/Library/Logs/Claude/mcp*.log
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.