IndiaRestros

What IndiaRestros MCP server does

The IndiaRestros MCP server gives AI agents access to a large directory of restaurants and food businesses across India. The dataset includes restaurants, cafes, bakeries, sweet shops, bars, dhabas, food courts, cloud kitchens, and street stalls in 8,911 cities and towns.

Restaurant records can include a permanent identifier, address, map point, phone number, opening times, current open status in IST, cuisines, cost for two, and amenities. Menus may contain sections, individual items, rupee prices, and vegetarian marks when those details are known. A canonical Indian dish catalog connects dish names across menus, enabling searches such as vegetarian thali options below a specified price.

The service also publishes source and observation-date information for facts, along with a payment-free ranking formula. Reads are free and do not require an API key or account.

How it works

The hosted endpoint is available at https://indiarestros.com/mcp and uses Streamable HTTP. Add that URL to an MCP client that supports remote servers. The repository is a listing and configuration repository; it does not contain software that must be installed or self-hosted.

The IndiaRestros MCP server handles read requests without authentication, with a documented limit of 120 requests per minute per IP. Optional account-oriented features use https://indiarestros.com/mcp/agent and OAuth. These features support saving restaurants, recording visits, and viewing a personal timeline.

Agents can combine restaurant, place, and dish queries to narrow results by location, cuisine, business type, availability, price, or menu item. The available data varies by listing, so a menu, phone number, or other field may not exist for every restaurant.

Setup and configuration

Connect by entering this URL in the MCP client's remote-server configuration:

https://indiarestros.com/mcp

For Claude Code, the documented command is:

claude mcp add --transport http indiarestros https://indiarestros.com/mcp

JSON-configured clients can use a server entry with the same URL. VS Code uses an HTTP server entry in .vscode/mcp.json. Gemini CLI can install the repository as an extension. Claude clients and ChatGPT provide connector settings where the endpoint can be pasted. No environment variables, API key, local package, or self-hosted process are required for the read-only connection.

Tools and capabilities

The IndiaRestros MCP server exposes 14 tools in these groups:

• Restaurant lookup: search_restaurants, get_restaurant, and get_menu
• Dish discovery: search_dishes
• Place browsing: list_places and get_place
• Data quality: report_correction and about_dataset
• OAuth account tools: whoami, set_handle, save_restaurant, list_saved, log_visit, and my_timeline

Typical tasks include finding a pure-vegetarian restaurant open now, retrieving a restaurant's phone number, comparing cost-for-two values, locating a dish below a rupee budget, and checking menus with prices. Place tools can expose states, cities, and areas with counts by cuisine and business type.

Limitations and notes

The IndiaRestros MCP server is a hosted service operated by IndiaRestros, so availability and data freshness depend on that service. Menu details and other attributes are provided where known, and each fact includes source and observation-date information. Read access is free, but the service documents a 120-requests-per-minute-per-IP limit.

The optional foodie tools are not part of the unauthenticated read flow; they use the /mcp/agent endpoint and OAuth. Commercial use is described as free for reads with attribution to the restaurant URL and source. The terms prohibit bulk resale of contact details. The repository files are released under the MIT license, while service use is governed by the linked IndiaRestros terms.