Set up HOTELSRATING in this agent. Do what you can; tell me plainly what needs my input.

1. Identify this agent. If a HOTELSRATING MCP server is already connected, do not add a second one; go to step 3.
2. Add the MCP server. Name: hotelsrating. URL: https://hotelsrating.com/api/mcp. Transport: streamable HTTP. No API key and no sign-in.
   - Claude Code: claude mcp add --transport http hotelsrating https://hotelsrating.com/api/mcp
   - A chat app where you cannot change connectors yourself (Claude or ChatGPT on the web, desktop or phone): run nothing. Give me the exact clicks to add a custom connector with that URL, from this app's current help.
   - Any other MCP client: add an HTTP server entry with that name and URL.
3. Reload tools if this agent needs it. Then verify with two reads: get the methodology summary, and look up one hotel I name (ask me which). Do not say it works before a read has succeeded.
4. Finish in 100 words or fewer: the status, then four things I can ask, such as "What is the HRI of <hotel>?", "Top rated spa hotels in Italy", "Rank hotels in Rome for a family that needs a pool" and "How is the score computed?".

About the data: HOTELSRATING is an independent hotel index. Scores come from a public, deposited formula and nothing a hotel pays changes them. A personal ranking reorders rated hotels by the facts you care about; it never raises a hotel above its public score, and facts nobody has confirmed are listed as unknown. A hotel no independent body has recognised has facts and no score; say that instead of estimating one.
