engagement

ChatGPT

Report a product click engagement event. Called by the UI when a user clicks on a product card.

search

ChatGPT

Use this tool when the user expresses shopping or product discovery intent. Examples include: - Searching for products (e.g., "black dresses", "running shoes") - Browsing categories (e.g., "show me jackets") - Applying filters such as color, size, price, brand, availability, or use case - Comparing products - Asking for product-specific information Guidelines: - Derive a structured search query from the user's intent, including category, attributes, constraints, and context. - If the request is too vague, ask a clarification question instead of calling the tool. - Do not generate, assume, or enrich product data. Use only the tool response. - Prefer calling the tool directly without unnecessary preamble when the user intent is clear. Price Display Guidelines: - Merchants may use different attribute names for pricing. - When product data contains both: - a current/active selling price, and - an original/reference/list price, display the current selling price as the primary price and the original/reference/list price with a strikethrough style. - When only a single price value exists, display that price normally. - Never rely on specific field names such as sale_price, discount_price, special_price, list_price, compare_at_price, or similar naming conventions. - Determine pricing display based on the semantic meaning of the values returned by the tool. - Only use pricing information explicitly returned by the tool response. - Do not calculate, infer, or generate discounts, savings amounts, discount percentages, price ranges, or any other price-related values. - If pricing information is unavailable or unclear, avoid discussing price. Assistant Response Guidelines: - Do not repeat the same message or product summary already returned by Shopping Muse assistant text. - Treat the Shopping Muse assistant text as the main product explanation. - Any additional chat response should add value beyond the Shopping Muse text. - Useful additional value can include: - Price-based recommendations using only prices visible in the widget - Simple comparisons between displayed products - Highlighting the best-value option among visible results - Suggesting how to refine the search, such as by size, budget, brand, color, or occasion - Calling out tradeoffs that are clear from the visible product data - Do not restate all product names, descriptions, or attributes already visible in the product widget. - If there is no meaningful additional value to add, keep the response minimal. Behavior: - Keep responses concise and avoid explaining the search process. - Do not repeat the same explanation before and after tool calls. - When results are displayed in the UI, such as an iframe or product widget, keep the pre-widget text minimal. - Since a text response may also appear under the product widget, avoid duplicating product details already visible in the UI. - In the post-widget response, refer only to products displayed in the widget. - When mentioning price, use only prices visibly displayed in the widget. - If both a current selling price and an original/reference price are displayed, refer to the current selling price as the current price and the struck-through price as the original price. - Do not calculate or infer discounts, savings percentages, or other pricing values. - Do not mention any price information that is not visibly displayed in the widget. - Prefer a short recommendation, comparison, or refinement suggestion after results are shown. This tool returns structured product results and may internally use search, recommendations, or personalization services.

Belk

Overview

Tools

engagement

search

Capabilities

Links