QuickReturns has an MCP server — a secure endpoint that lets an AI assistant read your store's returns data and answer questions about it in plain language. Connect it once, then ask things like "how many returns came in last week and what were the top reasons?" or "which products get exchanged most often?" without opening a dashboard.
Set it up
Everything you need is on one page in your QuickReturns dashboard:
app.quickreturns.ai/mcp-server
That page has your server URL, your store credentials, and the exact steps for the AI client you're using. Pick your client at the top and follow the steps shown — they differ slightly between Claude, ChatGPT and the rest.
Before you start
- You need Store Management access in QuickReturns.
- You need your store's API key and secret — find them under Settings → API Keys.
- You need an AI client that supports custom MCP connectors. Claude (web, desktop and mobile) and ChatGPT work today; Gemini, Cursor and others are rolling out. Any client that speaks streamable HTTP with OAuth will work.
Good to know
- Adding the connector and authorizing your store are two separate steps. If your assistant says it can't see your store, the connector was added but never authorized — go back to the MCP Server page and finish the second part.
- ChatGPT hides custom connectors by default. Turn on developer mode in ChatGPT's settings first, or the option to add one won't appear.
- Regenerating your API key ends existing connections. Remove the connector in your AI client and add it again afterwards.
Keeping your store safe
Treat your API key and secret like a password — anyone holding them can connect an assistant to your store data. Regenerate them under Settings → API Keys if you think they've been shared, and remove the connector from any AI client you no longer use.