Free tool · No sign-up
Can AI agents already act on your website?
Some platforms ship a Model Context Protocol endpoint by default. If yours does, AI agents can query your site — and in some cases book through it. This finds out.
What MCP is
A machine door into your website
The Model Context Protocol is a standard way for an AI agent to call tools rather than scrape pages. A site exposing an MCP endpoint is telling agents: here is a structured way to ask me things, and in some cases to do things.
That is a genuine channel. An agent can answer a customer's question from live content instead of a stale crawl, and where the endpoint allows it, complete a booking without the visitor ever opening your site.
It is also a decision most owners have not knowingly made. On Wix it is on by default and
advertised in an auto-generated /llms.txt — which is how we found the
one on this site.
The part worth checking
Reading is not the same as acting
Tools that only read — business details, site search — are straightforwardly useful. Tools that mint a visitor token and call your site APIs are a different category: they let an agent transact on a visitor's behalf.
Neither is wrong. Agentic bookings may be exactly the channel you want. But the two should not be confused, and this tool separates them rather than listing endpoints without comment.
MCP · Questions
About agentic access
What is an MCP endpoint?
A Model Context Protocol endpoint is a structured interface an AI agent can call instead of scraping your pages. It advertises a list of tools — things the agent is allowed to ask for — and the agent calls them directly. It is the emerging standard for agent-to-website interaction.
Does my site have one?
Run the check. If your site is on Wix, very likely yes: Wix serves one at /_api/mcp and advertises it in an automatically generated /llms.txt. Most other platforms do not, and a site with none loses nothing by it.
Can an AI agent really book an appointment through it?
Where the endpoint exposes a visitor-token tool alongside a call-API tool, yes — that is its documented purpose. The agent creates a visitor session and calls the same booking APIs your own site uses. Whether that is desirable is a business decision, but it should be a decision.
Is this a security problem?
Not inherently. These endpoints expose public information and act with visitor-level permissions, not admin ones. The risk is not a breach — it is running a channel you have never thought about, with no analytics on it and no idea what agents are doing there.
How is this different from llms.txt?
llms.txt is a static text file that AI crawlers overwhelmingly ignore — 408 fetches across 500 million AI bot visits in a 90-day window. An MCP endpoint is a live interface agents actually call. One is a suggestion; the other is a door.
Want this set deliberately?
Whether AI agents can read your site — and whether they can transact on it — should be your call. We can audit what is currently exposed and configure it to match what you actually want.