Skip to main content
Storefront chat uses the same public catalog search service as /api/v1/search, /api/v1/merchants/{merchantId}/search, /api/catalog, and the omni_search_catalog MCP tool. Chat is scoped to the deployment’s organization and excludes drafts and private seller data. A category or query is optional. The query also searches current public ACO FAQ questions and answers. Results include faqs, grouped by language with question, answer, and rank. Rejected FAQs and historical profiles are excluded. Variants use their own ACO profile when present and otherwise inherit the parent’s current profile, matching the public product page. For example, {"query":"dual monitor"} can find a product whose FAQ covers dual monitor compatibility even if its title does not. With sort=relevance (the default), title and SKU matches score 3; description, brand, and category matches score 2; and FAQ matches score 1. Scores add across matching fields, so a FAQ-only match ranks below any of these catalog fields. Matching multiple FAQs or marketplace categories does not multiply their score. Explicit newest, price, and rating sorts use their selected metric instead. Brand and category filters can combine with these arguments. Search supports every platform in OmniCommerce’s canonical marketplace list. Variant option filters are exact, case-insensitive matches; all requested options must match the same variant. Pagination uses limit and offset. newest sorts by catalogAddedAt, the date the catalog record was created, independently of later edits. It does not establish a release date or physical stock arrival. addedAfter is an inclusive timestamp with a time zone. Price ranges and price sorting use catalogPrice, the listed catalog price, and require a three-letter currency. Bounds are inclusive and may be zero. Products without a catalog price cannot match a budget. Prices with unknown or different currencies are excluded when a currency is selected. Marketplace offers remain separately labeled by currency and may have different prices. There is no implicit currency conversion or global cheapest ranking across currencies. Ratings average the imported, visible, valid 1–5-star product reviews for the product family. A marketplace filter also scopes the rating to that platform. The result includes rating.average, rating.reviewCount, rating.scale, and rating.source. Missing ratings return null and do not match minRating. Review requests, hidden reviews, invalid ratings, and private buyer details are excluded. These are imported review aggregates, not guaranteed lifetime marketplace totals. most_reviewed ranks review count, not sales or best sellers. Storefront results include review coverage limitations from the shared marketplace capability service. Shopee and Lazada currently have review read integrations. TikTok requires enablement, Shopify Shop reviews have no supported integration, and Zalora and eBay lack a confirmed product-review API in this integration. Amazon supports review requests rather than importing individual product reviews. WhatsApp has no product reviews. Catalog discovery and purchase links remain available for all these platforms. Unknown inventory remains inStock: null; it matches neither an in-stock nor an out-of-stock filter. For HTTP search, use min_price, max_price, min_rating, min_review_count, added_after, variant_options, and in_stock. variant_options is a URL-encoded JSON array. Invalid ranges, ratings, options, or price requests without a currency return HTTP 400.