Maybelline NY Makeup Try On

Records that the user accepted the virtual try-on Terms & Conditions for this session. Called by the widget automatically - do not call this tool directly.

accept_consent

Tool make new session for this conversation. Call ONCE, very start of every new conversation, before any other tool. After call, keep handling user original request—for example, call list_products if they asked about products, or ask what they want to try on.

create_session

Returns all shades in the same product line as the given UPC, in catalog order. Called by the widget automatically — do not call this tool directly.

get_product_siblings

Returns the VTO calibration data for a single product UPC. Called by the widget automatically when entering virtual try-on — do not call this tool directly.

get_product_vto_data

Use tool when need recommend product, get list available products. NEVER recommend product outside list returned by this tool. Tool lists products for virtual try-on across multiple makeup categories. Optional brand: pass exactly one slug from YSL, Maybelline, or omit brand to include all supported brands. Optional category: pass exactly one slug from blush, brow, concealer, contour, eyeliner, eyeshadow, foundation, highlight, lipcolor, lipliner, mascara, or omit category to include all supported categories. Do not pass multiple categories or values outside this list. If user says lipstick or lip color, filter or interpret as category lipcolor. Use the text result (VTO support + GEO descriptions when present) to choose products. ALWAYS show recommendations with product_carousel (do not dump the full catalog as chat text). After product_carousel, reply briefly in a conversational tone with two parts: (1) Open with a personalized lead-in that ties the picks to what the user asked for (for example: "Because you mentioned you want a bold matte, I'd recommend…"). (2) Then give one short bullet per shown product: bold product or shade name, em dash, and one concise sentence on why it fits (finish, color, wear, or the need it covers). Use product descriptions from the tool result when available. Do not invent product facts. Keep the carousel as the primary visual showcase.

list_products

Use this tool whenever you need to show product information to the user. This tool opens an interactive product carousel widget with a full virtual try-on flow. The input parameter is named "upcs" (NOT "products") — pass an array of 1-5 UPC strings. UPCs are strings: copy the exact "upc" string value from list_products output, preserving all leading zeros (e.g. "0041554429893" NOT "41554429893"). Never treat a UPC as a number. The upcs MUST come from a call to list_products tool. NEVER call this tool with products that have not been retrieved from list_products tool. All UPCs must exist in the catalog - use exact "upc" string values from list_products, never guess. After product_carousel, reply briefly in a conversational tone with two parts: (1) Open with a personalized lead-in that ties the picks to what the user asked for (for example: "Because you mentioned you want a bold matte, I'd recommend…"). (2) Then give one short bullet per shown product: bold product or shade name, em dash, and one concise sentence on why it fits (finish, color, wear, or the need it covers). Use product descriptions from the tool result when available. Do not invent product facts. Keep the carousel as the primary visual showcase.

product-carousel

Records a widget UI analytics event. Called by the widget automatically — do not call this tool directly. sessionId and userId are taken from ChatGPT request _meta on the server, not from tool arguments.

record_widget_event

Registers a selfie captured inside the product carousel widget. Called by the widget automatically after the user takes a selfie in the selfie screen — do not call this tool directly. Do NOT ask the user to upload a selfie or call this tool yourself; the widget handles the entire capture and upload flow.

set_uploaded_selfie