Cara

Retrieves detailed information for a specific VIN (Vehicle Identification Number). Returns comprehensive vehicle specifications including engine, transmission, drivetrain, MPG, features, and photos. Expanded view enables full screen view and includes payment calculator and dealer details viewer. RESPONSE STRUCTURE: • structured_content returns KEY FACTS ONLY: vin, year, make, model, trim, price, miles, body_type, exterior_color, drivetrain, fuel_type • The widget displays comprehensive vehicle details to the user including: - Multiple vehicle images (if available) - Full feature list - Options list - Seller comments - Payment calculator - Dealer information • DO NOT call this tool again to "see more" - the user already sees everything in the widget • One successful call provides complete vehicle information

get-vin-details

Searches U.S. vehicle inventory using filter conditions with operators and preference levels. RESPONSE STRUCTURE: • structured_content returns SUMMARY DATA ONLY: total count, price/mileage ranges, top makes/models, pagination cursors • The widget displays full vehicle listings to the user (you do NOT receive individual vehicle details) • DO NOT call this tool again to "get more details" - the user already sees the vehicles in the widget • One successful call is sufficient - present the summary stats and let the user browse the widget ZERO RESULTS HANDLING: Note: If search tool gets 0 results, it will remove semantic tags and retry automatically. If still 0 results, it will return 0 results to you. If search returns 0 vehicles, retry by relaxing filters: 1. Broaden search radius (e.g., +50 miles) 2. Remove or relax conditions with lowest preference_level first (e.g. widen numeric ranges (price, miles, year) by 20-30%) 3. Inform user what filters were relaxed KEY RULES: • U.S. vehicles only • Extract ALL user criteria into conditions array with: field, operator, value, preference_level • Operators: Numeric='=','<=','>=','<','>'; String/Enum='in','not_in' (values always lists) • For ranges: use '>=' for min, '<=' for max as separate conditions PREFERENCE LEVELS (1-5) - based on user language: • 5: "must have", "required", "only" → non-negotiable (use sparingly) • 4: "important", "really want" → strong preference • 3: "prefer", "would like" → moderate preference • 2: "would be nice", "looking for" → mild preference • 1: "not important", "would be a plus" → minor preference • Default when no emphasis: budget=4, make/model/body_type=3, color/features=2 • Higher preference_level = greater impact on fit_score ranking GUIDELINES: • Inference OK: "Japanese makes" → make in ['Toyota','Honda','Nissan','Mazda','Subaru','Mitsubishi'] • Negation OK: "not white" → base_exterior_color not_in ['White'] • Use body_subtype ONLY for Pickup/Chassis Cab body_types • Avoid contradictory filters: don't combine mutually exclusive options (e.g., has_leather_seats AND has_leatherette_seats, awd AND fwd, etc.) • Group features and semantic_tags into single conditions into their own lists REQUERY INSTRUCTIONS: • If prior search returned 0 results retry with new conditions • If user changes ANY filter criteria or sort order, call this tool again with updated parameters • This tool does not support in-place filter/sort modifications—always requery with the full updated conditions

search_vehicles