1 command setup Add Weavify to Claude in seconds with a single terminal command.

7 MCP tools Four discovery tools that answer today, and three booking tools that decline while reservations are closed.

126 venues Restaurants, bars and nightclubs across Bangkok. Bangkok only.

AI assistants can recommend restaurants all day long, but acting on the recommendation means leaving the chat window. Weavify's MCP server closes the first half of that gap: Claude can search the directory, read a venue's menu, hours, address and phone number, and hand you the details without you opening a browser. The second half, holding a table, is built and not yet open. This guide says exactly which calls answer today.

Booking status, as of September 2026. No venue on Weavify currently accepts reservations through Weavify. The three booking tools below are implemented and respond, but every venue returns BUSINESS_NOT_REGISTERED together with its own phone number and website, and availability search returns no slots. Use the discovery tools, then contact the venue directly. Booking is a per-venue setting, so this may change for individual venues.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI models connect to external tools and services. Think of it as giving Claude hands, instead of just talking about restaurants, it can actually interact with Weavify's booking system to search businesses, check table availability, and make reservations.

MCP is supported by Claude Desktop, Claude Code, Gemini CLI, Cursor, Windsurf, and a growing ecosystem of AI clients. Weavify's MCP server speaks Streamable HTTP at https://weavify.io/mcp, so it works over a plain HTTPS connection with no special setup. The older SSE transport at /mcp/sse still responds for clients that need it.

Step 1: Add Weavify to Claude

Option A: Claude Code (Terminal)

If you use Claude Code, run this single command:

claude mcp add --transport http weavify https://weavify.io/mcp

That's it. Claude now has access to all 7 Weavify tools.

Option B: Claude Desktop

Open your Claude Desktop configuration file (claude_desktop_config.json) and add Weavify to the mcpServers section:

{
  "mcpServers": {
    "weavify": {
      "type": "http",
      "url": "https://weavify.io/mcp"
    }
  }
}

Restart Claude Desktop and you'll see the Weavify tools appear in the tool picker. No API key, no OAuth, no authentication of any kind required.

Step 2: Ask Claude to Find a Business

Now just talk to Claude naturally. Here are some example prompts:

"Find me a rooftop bar with good cocktails in Thonglor for 4 people tonight around 8 PM."

"I need a fine-dining Japanese restaurant in Sukhumvit for a business dinner, 6 people, Friday night."

"What are the best speakeasy bars in Silom? Do any have availability for 2 people this Saturday?"

Claude will call the search_businesses tool behind the scenes, searching across cuisine, ambiance, location, and price range to find the best matches.

Step 3: What happens when you ask Claude to book

Once Claude presents venue options, ask it to book and it will try the booking chain. Here is what each call returns today:

Call check_availability, which returns an empty list for every venue while reservations are closed
Call create_booking_hold, which returns 422 BUSINESS_NOT_REGISTERED along with the venue's own phone number and website
Stop there, because confirm_booking has no hold to confirm
Hand you the venue's contact details so you can call the restaurant yourself

So the part of the flow that completes today is search, read the profile, hand off. That is two calls and a phone number, without a browser tab. Do not tell a guest a table is held.

The 7 MCP Tools

Here's what Claude can do once connected to Weavify:

search_businesses Search by location, cuisine, ambiance, price or free text across the 126 published venues

get_business_details Full venue profile: menu with prices, opening hours, address, phone number and website

check_availability Queries bookable slots for a date and party size. Returns an empty list for every venue today

create_booking_hold Places a 10-minute hold on a table. Declines with the venue's contact details while reservations are closed

confirm_booking Confirms a hold into a booking and returns a WV-XXXXXX code. Unreachable while the hold call declines

get_booking_status Looks up a booking by its WV-XXXXXX confirmation code

cancel_booking Cancels a confirmed booking and releases the inventory

Not Using Claude? Other Options

Weavify's MCP server works with any MCP-compatible client:

Gemini CLI: Connect using the MCP server URL https://weavify.io/mcp
Cursor / Windsurf: Add as a remote MCP server in settings
ChatGPT: Use our OpenAPI spec to create a Custom GPT Action
Direct API: REST endpoints at https://weavify.io/api/v1, see the full documentation

Weavify publishes its data over MCP, an OpenAPI 3.1 description and plain REST, so any client that speaks one of those can read it. We have not tested Weavify against Google and Shopify's Universal Commerce Protocol and make no claim about it.

Why This Matters for Businesses

Every time a traveler or local asks Claude "where should I eat in Bangkok tonight?", that's a potential customer. If your venue is on Weavify, Claude can answer that question with your menu, your hours and your phone number instead of a guess. Reservations through Weavify are a separate switch, per venue, and it is off everywhere today.

This is what we call the shift from search to execution. Discovery is only half the equation. The businesses that let AI agents complete the transaction are the ones that will capture the agentic web's demand.

If you run a Bangkok restaurant, bar, or nightclub, claim your venue on Weavify to publish your details where compatible assistants can read them. Being listed does not guarantee that any assistant will recommend you.