# booking.forktime.ai

> Restaurant reservations served by Forktime AI.
> This origin is multi-tenant: each restaurant lives under its own path.
> There is no public index of restaurants on this origin.

## Reaching one restaurant

- Booking page: https://booking.forktime.ai/r/<restaurant-slug>
- Plain-text summary (hours, address, phone, how to book): https://booking.forktime.ai/r/<restaurant-slug>/llms.txt
- Structured menu (JSON-LD): https://booking.forktime.ai/r/<restaurant-slug>/menu.json
- Live walk-in waitlist: https://booking.forktime.ai/w/<restaurant-slug>

The slug is the restaurant's identifier in any Forktime link you already
have. If you do not have one, ask the restaurant for its link.

## Booking programmatically

Prefer the API over scraping the booking page.

- MCP server (Streamable HTTP): https://api.forktime.ai/mcp
- Authorization server metadata: https://api.forktime.ai/.well-known/oauth-authorization-server
- Protected resource metadata: https://api.forktime.ai/.well-known/oauth-protected-resource

The MCP endpoint requires OAuth. An unauthenticated request returns 401
with a `WWW-Authenticate` header naming the protected-resource metadata,
so the discovery flow is standard (RFC 9728) — start there.

## Not for crawling

Guest reservation links (`/track/<code>`, `/i/<code>`) are capability
URLs: the code is the credential. They are disallowed in robots.txt and
must not be indexed or shared.