About nanocart-mcp
nanocart-mcp is NanoCart’s official MCP server for AI agents managing a store. Its tools cover products, orders, coupons, subscribers, settings, and sales reports, with destructive actions marked for client-side confirmation.
Key features
- 01Manage products, variants, and images
- 02Handle orders and shipping details
- 03Manage coupons, categories, and subscribers
- 04Review sales reports and update settings
- 05Mark read-only and destructive tools
Use cases
- 01List and update store products
- 02Mark orders shipped with tracking details
- 03Create coupons with discount conditions
- 04Review recent sales and orders awaiting shipment
MCP support
It offers local stdio and hosted Streamable HTTP MCP access, exposing 39 tools for managing a NanoCart store.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://mcp.nanocart.io/mcpclaude mcp add --transport http bytebunny-io-nanocart https://mcp.nanocart.io/mcp{"mcpServers":{"bytebunny-io-nanocart":{"type":"http","url":"https://mcp.nanocart.io/mcp"}}}Open in Cursor{"servers":{"bytebunny-io-nanocart":{"type":"http","url":"https://mcp.nanocart.io/mcp"}}}Open in VS Code{"mcpServers":{"bytebunny-io-nanocart":{"type":"http","url":"https://mcp.nanocart.io/mcp"}}}{"mcpServers":{"bytebunny-io-nanocart":{"serverUrl":"https://mcp.nanocart.io/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://mcp.nanocart.io/mcpAuth requiredTools
Details
From the project READMEnanocart-mcp
Official NanoCart MCP server — run your store from any MCP-capable AI agent (Claude Code, Claude Desktop, Cursor, Codex, and friends). List and update products (with variants and images), manage orders, coupons, categories, subscribers and settings, and pull sales reports — by asking.
39 tools. Reads are marked read-only; destructive tools (archive product, delete coupon) are annotated so your agent asks before acting. Refunds, billing changes, and API-key management are deliberately NOT exposed — do those in your dashboard.
Credentials
From your dashboard: Settings → API Keys (sc_live_..., keep secret) and
Settings → Store Information (your store ID). The key grants full admin
access to your store — treat it like a password.
Option A — Local (recommended)
Claude Code:
claude mcp add nanocart --env NANOCART_API_KEY=sc_live_... --env NANOCART_STORE_ID=your-store-id -- npx -y nanocart-mcp
Any client that speaks stdio MCP (JSON config style):
{
"mcpServers": {
"nanocart": {
"command": "npx",
"args": ["-y", "nanocart-mcp"],
"env": {
"NANOCART_API_KEY": "sc_live_...",
"NANOCART_STORE_ID": "your-store-id"
}
}
}
}
Option B — Hosted (no install)
Endpoint: https://mcp.nanocart.io/mcp?store=your-store-id with header
Authorization: Bearer sc_live_... (Streamable HTTP).
Claude Code:
claude mcp add nanocart --transport http "https://mcp.nanocart.io/mcp?store=your-store-id" --header "Authorization: Bearer sc_live_..."
Note: web clients that only support OAuth-based custom connectors (e.g. claude.ai's connector UI) can't send API-key headers yet — use Option A there.
Example prompts
- "List my draft products and publish the ones with images"
- "Create a SUMMER20 coupon — 20% off, min order $50, expires end of August"
- "What did I sell last week? Any orders I still need to ship?"
- "Mark order NCT-1042 shipped with tracking 9400 1000 0000 0000 0000 00"
- "Upload the photos in ./shots and create products from them at $24 each"
Safety
- Secrets are redacted from every response; your key is never echoed or logged.
- Destructive tools carry MCP
destructiveHintannotations — well-behaved clients prompt before running them. - Prices are explicit
price_centsintegers (no dollars/cents ambiguity).
Pairs with the NanoCart skill (integration knowledge for building the cart into websites). Docs: https://docs.nanocart.io/#ai-build · AI-readable: https://docs.nanocart.io/llms-full.txt
MIT · © 2026 NanoCart · a ByteBunny, LLC company
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.