Gondola MCP Server

The Gondola MCP Server is the official, continuously reachable remote service offered by Gondola for AI-assisted travel search. The product page describes a standalone MCP integration that works with Claude, ChatGPT, Cursor, Windsurf, VS Code, and other compatible clients. The public service is operated at mcp.gondola.ai; for interactive connections, Gondola documents an authenticated MCP endpoint. Its identity is therefore not just a directory record: it is tied to a dedicated product page, provider documentation, and a provider-operated endpoint intended for ongoing use. The reviewed provider sources do not identify a separate public source repository for this deployed service.

Travel information in an AI client

According to the provider, the server can search hotels, flights, and rental cars. For hotels it returns current availability, room rates, cancellation terms, and member or points information. Its distinctive focus is comparing cash prices with loyalty-point redemptions. This lets an AI client search by destination and dates while also explaining which option may offer better value relative to the points used. Other documented tools provide hotel details, reviews, vehicle details, insurance information, and booking links. Flight support centers on route searches and a link for completing the booking.

The results are decision support, not a guarantee of availability, final price, or a particular hotel benefit. Travelers must review terms, taxes, fees, cancellation rules, and the actual checkout flow before booking. An AI agent should not turn a search response into a binding promise. A safer workflow begins with read operations, asks the traveler to confirm the selected option, and then opens the checkout link returned by Gondola. According to the provider, in-conversation booking is limited to approved booking partners; for ordinary integrations, the checkout link is the intended path.

Connection and durable product boundary

Gondola documents a standards-based OAuth-protected MCP server with protected-resource metadata, an authorization server, and dynamic client registration. An MCP client should discover the published metadata and use the authorization-code flow with PKCE. Credentials, tokens, client secrets, and connection strings do not belong in a catalog description and are not stored here. A public discovery endpoint supports directory listing without authentication. Tool calls, however, require a Gondola sign-in and an account that has completed onboarding. This separation between public product discovery and authenticated use makes the service recognizable as an independent remote product.

Data, roles, and write access

Authorization uses Gondola OAuth. The provider documents a read scope for travel and account data and a separately requested write scope for price alerts. A booking scope is available only to approved booking partners. Clients therefore cannot freely expand their role or write access. Even so, read access may expose personal travel profiles, loyalty accounts, point balances, past trips, saved payment information, or itinerary data. Write capabilities can create or delete rate alerts. Booking and cancellation functions, when enabled for a partner, can have especially consequential effects.

Gondola processes data from travel and account contexts through these tools. Which tool results are forwarded to a language model, logged, or processed further also depends on the connected AI client and its provider. Before using the integration with company or personal data, review authorization, retention, regional processing, privacy terms, and onward-sharing paths for that client. Hotel, review, and profile content must be treated as untrusted input because it can try to steer an agent into unwanted actions. Use the minimum necessary permissions, confirm bookings outside the model, and review every change to alerts or travel information.

Assessment

Gondola is an independent provider-operated remote MCP server rather than a short-lived demo script or an implementation detail inside an unrelated project. The reviewed product page names the service, describes its tools, and links to dedicated OAuth integration documentation. Its durable identity remains tied to Gondola and its published endpoints. The catalog information was reviewed against these official sources on September 9, 2026.

MCP-compatible client, a Gondola account with completed onboarding, and OAuth 2.1 with PKCE support. Write and booking scopes are separately restricted.

Add the official Gondola MCP Server to the MCP client through the provider product page. For custom clients, use the documented discovery endpoint, retrieve OAuth metadata, and complete the authorization-code flow with PKCE. Do not share tokens or secrets in configuration files.

OAuth 2.1 with authorization code and PKCE. Tool calls require a Gondola sign-in and completed onboarding; public discovery remains anonymous for registries.

mcp:read is the default for search and the member’s own travel data. mcp:write requires explicit approval and covers rate alerts. mcp:book is available only to approved booking partners.

Tools can process travel profiles, trips, loyalty accounts, point balances, and search data. The connected AI client additionally determines which results are forwarded to a model.

Personal travel and account data may be exposed. Write access to rate alerts and approved booking actions should be confirmed outside the model. Hotel and review content may contain prompt injection.