FridgeBase in Claude, ChatGPT and AI assistants

The MCP server's address is https://fridgebase.com/mcp. In Claude, add it as a custom connector. In ChatGPT, add it as a connector. Once it is added, you ask your question in everyday words and the assistant calls the tools it needs.

• search_models: find models by need in plain words and by filters.
• get_model: all the data of one model.
• compare_models: compare two to four models.
• list_filters: the filters and what they mean.
• list_guides: the buying guides.
• search and fetch: the search ChatGPT uses, deep research included.

The assistant gets our data and the official data with its source. It never gets the Amazon price, title, image or rating. For those, it sends you to the model's page on FridgeBase, which shows them up to date.

The same data goes through a REST API described at /openapi.json, which you can use for a custom GPT's actions. The full catalog is at /catalog.json and the site's summary for AI tools at /llms.txt.

These requests use filters that already exist:

• Show me fridges that fit in a small kitchen.
• Which bottom-freezer fridges are most efficient for their size?
• Compare the quietest models I can find in a standard width.
• Find French-door fridges with a water dispenser.
• What are the best-rated models with at least a hundred customer reviews?

The same server also answers for France, with Amazon.fr and fridgebase.fr, in French. Every tool takes a market parameter: us or fr, us by default. Each market has its own catalog in its language, and each model's address leads to its market's site.

Our data comes from Amazon, the U.S. Department of Energy and ENERGY STAR. The official figures come from the U.S. Department of Energy's Compliance Certification Database and ENERGY STAR. If something does not match, write to us through the contact page.