Yes. The Gemini CLI supports MCP servers through the gemini mcp add command, over both local (stdio) and remote (HTTP) transports. That covers both forms of Guesty's official MCP server (beta): the local npx -y @guestyorg/sdk mcp command, and the hosted endpoint at https://mcp.guesty.com/v1 with a bearer token header.
What works today
Guesty's documentation lists Cursor, Claude Desktop, and VS Code as its tested clients, but the local server speaks standard stdio MCP, which is exactly what the Gemini CLI consumes from ~/.gemini/settings.json. The server is read-only during beta, so Gemini can look up and summarize your Guesty data but cannot change records.
The same goes for a server you build yourself against the Guesty Open API: it plugs into the Gemini CLI and Claude Code identically, and you control exactly which tools it exposes. Your OAuth client ID and secret stay in environment variables on your own machine either way.
The hosted API route
If you would rather call an API than run and maintain a server, there is a hosted middle path: the RapidEye API exposes Guesty as a hosted connector, so the credential handling and token care above happen on our side and your assistant or script calls one consistent surface. The catalogue is machine-readable, which means an AI agent can discover it and hand you the next step on its own.