Every OpenBooking business has an MCP address an AI assistant can connect to: https://app.openbooking.sh/b/{id}/mcp on the hosted service, or /mcp on a backend you run yourself. Add that address in Claude, ChatGPT or Cursor and ask it to book. The assistant shows real times, prices and terms, and books only after you say yes.
https://app.openbooking.sh/b/studio-nord/mcp
The address ends in /mcp. It is the same on every
client. The one above is the demo salon, so you can try it right
now.
https://app.openbooking.sh/b/{id}/mcp
Replace
{id} with the business’s URL name, for example
studio-nord. Every business’s booking page is
https://app.openbooking.sh/b/{id}.
https://your-domain.example/mcp
A backend from
npx create-openbooking serves MCP at
/mcp. Claude and ChatGPT need an HTTPS address they
can reach, so deploy it first.
https://app.openbooking.sh/mcp
One address for
every listed business: the assistant finds one with
find_business, then books it.
Menus change often, so each assistant’s own page is linked as the source of truth.
In Claude, go to Customize > Connectors, click Add custom connector, enter the MCP address and click Add. On Team and Enterprise plans an Owner first adds it under Organization settings > Connectors, then each member connects it from Customize > Connectors. Then start a chat and ask Claude to book.
Source: Claude: custom remote MCP connectors, checked 17 October 2026.
Go to chatgpt.com/plugins,
select the plus button, then Add custom MCP server. Give it a
name and description, choose Public endpoint under
Connection and enter the MCP address including
/mcp. Configure authentication as the screen asks
(OpenBooking needs no sign-in), accept the risk warning, then select
Create as a plugin. Install it, start a new conversation,
type @ and pick it. Account and workspace policies
apply, so your plan or admin may limit this.
Source: OpenAI: connect your MCP server to ChatGPT, checked 17 October 2026.
Cursor reads MCP servers from .cursor/mcp.json in your
project, or ~/.cursor/mcp.json for all projects. Add
the address as a remote server, then manage it from Cursor’s
Customize page:
{
"mcpServers": {
"openbooking": {
"url": "https://app.openbooking.sh/b/studio-nord/mcp"
}
}
}
Source: Cursor: Model Context Protocol, checked 17 October 2026.
Use the Streamable HTTP transport with the MCP address and no authentication. To check that an address works from a terminal, send an initialize request. A working server answers with its name and booking instructions:
curl -X POST https://app.openbooking.sh/b/studio-nord/mcp \
-H "content-type: application/json" \
-H "accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"test","version":"1"}}}'
Protocol: modelcontextprotocol.io.
It reads the business (get_business_info), searches free
times with price and cancellation terms
(search_availability), holds a slot
(hold_slot), and books it only after you agree
(confirm_booking). It can also look up, reschedule and
cancel a booking. Holds expire and a retried request never
double-books.
No. The booking endpoints answer without authentication. Customers are asked for their name and email when they confirm a booking.
Both connect from their own servers and need an HTTPS address they can reach. To try a local backend, deploy it (npx vercel) or use a tunnel as described in each assistant’s own documentation.
Streamable HTTP, the current remote transport in the Model Context Protocol. A plain POST with an initialize request returns the server’s name and instructions.
Yes. Connect the demo salon at https://app.openbooking.sh/b/studio-nord/mcp. It is a demo business, so don’t enter real personal details when confirming a booking.