Ford

Use this whenever a shopper wants to browse, explore, compare, or shop Ford vehicles: asking which models or body styles Ford offers (SUVs, trucks, cars, electric, commercial), "find me a truck", "show me electric SUVs", needs-based asks like a family vehicle or something that can tow, comparing models or model years, or what vehicles cost or fit a budget. Call it instead of answering from memory so the shopper gets the interactive vehicle carousel; when the intent is clear, call it right away with the closest category, price range, or name rather than asking clarifying questions first, then summarize the result instead of re-listing every vehicle. Browse the current public Ford en_us vehicle catalog, optionally filtered by category, exact vehicle name, MSRP range, model year, trim, minimum seating capacity, minimum towing capacity, or powertrain. Category, name, model year, MSRP range and trim filter results. Seating, towing and powertrain are preferences: they never exclude a vehicle, they set the display order (verified matches first, then vehicles with unpublished values, then vehicles that miss a preference) and list any missed ones in unmet_preferences; an unpublished capacity is unknown and never counted as missed. A name filter returns every matching model year, useful for a multi-year nameplate like Bronco. name and model_year also accept JSON arrays (up to 10 each) to fetch several vehicles or years the shopper asked about in a single call, e.g. {"name": ["Bronco", "Explorer"], "model_year": [2025, 2026]}; a vehicle matches if it has any listed name and any listed year. Use one whole nameplate per array element, never a comma-separated string. For a follow-up detail request, use the exact name field returned here; category can refine the next prompt. Do not use the internal id or model_id. ALWAYS also send vehicleConfigurationIntent, callerApp, and callerModel on every call.

get_all_models

Search new Ford vehicles in stock or in transit at U.S. Ford dealers near the shopper. Requires a Ford nameplate such as F-150, Bronco, or Mustang Mach-E. Location: pass the shopper's 5-digit ZIP code if they gave one, or location_ref from a previous result; otherwise the tool uses the approximate location shared by the app when available and labels results as approximate, or returns location_required so the shopper can enter a ZIP. Never invent a ZIP code. Optional filters: model year, trim, exterior color families, drivetrain, MSRP range, search radius (default 50 miles), and sort order (default nearest first). Returns up to 6 listings per page with year, trim, MSRP, estimated net price with its disclosure, availability, dealer name, city, state, and distance. Prices are dealer estimates and exclude taxes and fees.

search_inventory

Use this whenever a shopper asks about one specific Ford model: its price or starting MSRP, trims, specs, features, capability, or says things like "tell me about the Bronco" or "show me the Mustang Mach-E". Call it instead of answering from memory so the shopper gets the interactive model view; when the shopper names a model and a model year, call it directly with both; otherwise call get_all_models with that name first, because many models have several model years and the detail call needs the chosen model_year. Get details for one Ford en_us vehicle using its exact display name from get_all_models. If the name matches more than one model year, also pass model_year from the chosen catalog item. Never send id or model_id. ALWAYS also send vehicleConfigurationIntent, callerApp, and callerModel on every call.

get_model_details

Used by the inventory app to get current details for one Ford dealer listing the shopper selected from search_inventory results: whether it is still available, year, trim, MSRP and estimated net price, dealer, and up to 8 equipment highlights.

get_inventory_vehicle_details