A sensible starting point
“Find automatic petrol SUVs under ₹15 lakh ex-showroom. Sort by price and show five options.”
Uses search_cars. ₹15 lakh is 1,500,000 INR.
Bring ARSA’s car catalogue and expert discovery into an MCP-compatible assistant. Build a shortlist, compare the details, and find your next step—all through conversation.
A small setup. More useful answers.
MCP (Model Context Protocol) lets an assistant request information from ARSA’s tools. Your client must support remote MCP over Streamable HTTP; availability depends on its version, plan, and workspace settings.
Open your assistant’s connector or MCP settings. Choose its option for adding a custom remote server.
Name it ARSA Car Advisor and enter the server URL. Public tools need no account. For personalized recommendations, use your client’s OAuth account connection and approve access on ARSA.
Save the connection and enable ARSA for your chat if your client requires it. Try one of the prompts below and review any tool-use approvals your client presents.
Use a compatible client such as Claude, ChatGPT, Cursor, or Windsurf where remote MCP is supported. Account linking requires OAuth support in your client. Connector-directory approval is separate.
MCP server URL
https://arsa-production.up.railway.app/mcpExample JSON config
{
"mcpServers": {
"arsa": {
"url": "https://arsa-production.up.railway.app/mcp"
}
}
}For clients that accept remote servers using this schema. Client configuration formats vary; follow your client’s own documentation.
From research to a shortlist
“Find automatic petrol SUVs under ₹15 lakh ex-showroom. Sort by price and show five options.”
Uses search_cars. ₹15 lakh is 1,500,000 INR.
“Find the Hyundai Creta and Kia Seltos in the catalogue, then compare two specific variants on price, mileage, and power.”
Search first, then compare the returned car IDs.
“Find experts who specialize in electric vehicles, speak Hindi, and have a rating of at least four stars.”
Discover profiles; arrange consultations through ARSA.
These tools retrieve catalogue, expert, and pricing information. They cannot book consultations, make payments, or change your account.
Your preferences. Your permission.
Connect your existing ARSA account to get recommendations using your saved car preferences. Sign in on ARSA with your email and password, or request an email code—even if you normally use Google login. Then choose Allow connection.
The ten public tools work without an account. After you approve read-only access, your assistant can also use get_personalized_recommendations, with your saved preferences. It cannot change your account, book consultations, or make payments.
You can review and revoke connections at any time. Manage your saved preferences in the ARSA app.
Review ARSA’s privacy policy and terms before connecting. Your assistant provider’s policies also apply to your conversations and tool results.
Check that your client supports remote Streamable HTTP, the URL ends in /mcp, and ARSA is enabled in the current conversation. Public access should use no authentication.
The server is a protocol endpoint, not a website. Connect through an MCP client; opening the URL in a browser is not a reliable connectivity test. The transport is stateless and uses POST requests.
Yes, through an OAuth-compatible MCP client after the updated server is deployed. Sign in on the ARSA authorization page, review the requesting client, and approve arsa:read. Never put passwords or app access tokens in a conversation. Revoke access from the connections page.
Try a specific brand, a narrower price range, and a smaller result limit. If the problem persists, contact ARSA through the help center. Do not include access tokens or passwords in your report.