compare_products
ChatGPTCompare two to five named products side by side. Answers 'X vs Y', 'which of these is healthier or cleaner', 'should I buy X or Y' and 'rank these for me' when the user names the products. Each entry is a product name with its brand ('Oatly Oat Milk Original', 'Chobani plain Greek yogurt') or a barcode; each requires an exact product identity; ambiguous names return candidates for confirmation instead of an assumed score. Returns confirmed products sorted by score only when comparable, otherwise in requested order, each with its Scout score and label, verdict, the count of flagged ingredients with up to five of their names, the count of beneficial ingredients, NOVA processing group, hand-reviewed and lab-tested flags, available lab measurements with detection statuses and source, and page URL, and names the entries that could not be found, need confirmation or are still being analysed. A winner requires all entries to be distinct, confirmed, scored, non-provisional and in the same specific Scout category; ties are explicit and cross-category comparisons have no winner — the score, verdict and flagged ingredients behind a which-is-better-and-why answer are in the one response. Not for a product type with nothing named ('best oat milk' is get_category_rankings) or a single product (get_product_score).
find_higher_scoring_alternatives
ChatGPTFind higher-scoring swaps for one product. Answers 'healthier alternative to X', 'what should I buy instead of X', 'something like X that scores better' and 'is there a cleaner version of X'. Lookup by barcode or exact product name; incomplete names return candidates to confirm before suggesting swaps. Returns the product's own score and page URL, then up to eight products that score above it: up to four hand-reviewed picks from the product's Scout category, each scoring above the product and at least 70 — the same product type when the category holds one, otherwise from the wider shelf (for example popcorn for a tortilla chip), and the response labels which — and up to four catalog products near the same aisle scoring 80 or higher, labelled when they are not all the same type, each with score and page URL. Says so when the product has no score yet or when Scout has no higher-scoring product of that type. Not for 'the best X' with no starting product (get_category_rankings).
get_category_rankings
ChatGPTRank Scout's hand-reviewed products in one category from best to worst score. Answers 'healthiest X', 'best X', 'cleanest X', 'top-rated X', 'which X scores highest' and 'X without ingredient Z' for a type of packaged food, drink, supplement or personal care — for example 'protein powders', 'greek yogurt', 'oat milk', 'bottled water', 'sunscreen', 'baby formula'. Accepts everyday names: the input is matched to the closest Scout category, and extra type words narrow the list ('oat milk' ranks the Milk category's oat products; the response says when nothing matches the extra words). Returns the ranked products with rank, score and label, hand-reviewed, lab-tested and Scout's Pick flags and page URLs, the number of reviewed products in the category and the category page URL — scores and flags only, no verdicts or ingredient lists. Only hand-reviewed products are ranked, so a category holds from a couple of products to a few dozen. When the name matches no Scout category, the response lists the closest categories and the best-scoring catalog products whose names match the words. Not for a specific named product (get_product_score), a comparison of named products (compare_products) or restaurant food (get_chain_menu).
get_ingredient_info
ChatGPTExplain one food, drink, supplement or personal-care ingredient or additive. Answers 'is Z bad for you', 'is Z safe', 'what is Z', 'why is Z flagged' and 'should I avoid Z' — for example 'carrageenan', 'sodium benzoate', 'E621', 'high oleic sunflower oil', 'titanium dioxide'. Returns Scout's typical read of the ingredient (its sentiment), an overview, risks, benefits, perspectives (regulators, research, diets) each with a stance, dietary flags, common uses, up to three cited sources, how often it appears in Scout's hand-reviewed products with the positive, neutral and negative split and up to three example products with scores, and the ingredient page URL when one exists. Says so when Scout has no profile for the ingredient. Not for a product's own ingredient list (get_product_score).
get_product_score
ChatGPTGet Scout's full analysis of one product. Answers 'what does X score', 'is X healthy', 'what's in X', 'why does X score low' and 'what is this barcode'. Returns the score (1–100) with its label, Scout's verdict, category and category page URL, NOVA processing group, up to twelve key ingredients each marked positive, neutral or negative with the reason, the counts of flagged and beneficial ingredients, USDA Organic, seed-oil-free, Scout's Pick and lab-tested flags, published lab results (contaminant, original amount and units, level, detection status, source and testing notes) when Scout has them, and the product page URL. A barcode lookup is exact; a name lookup requires an exact product identity. Broad or incomplete names return needsConfirmation with up to five unscored candidate identities, identified by brand plus full name or barcode. Missing brand or variant words never fall back to a different product. Lab evidence explicitly distinguishes published panels, testing claims without a panel and unavailable results; missing dates and report URLs remain null. A product in the database whose analysis is still in progress comes back with no score and a note saying so. One product per call: two or more named products are compare_products, and 'the best X' with no product named is get_category_rankings.
get_scoring_methodology
ChatGPTExplain how Scout scores products, in general or for one category. Answers 'how does Scout score X', 'what does a Scout score mean', 'why is this score capped', 'do lab results change a score', 'does Scout get paid by brands' and 'what doesn't Scout score'. Returns Scout's published methodology, quoted from scouthealthapp.com/methodology: the 1–100 scale and its four bands, the four steps (ingredients judged against a framework written for the category, a processing cap from the published NOVA group, packaging deductions, and lab results shown as disclosure rather than scored), the fixed ingredient rules and post-scoring overlays that apply, the labels shown next to a score, what Scout does not score, and its independence statement. With a category it also names the framework that covers it where Scout has a dedicated one, and gives the number of hand-reviewed products, how many are lab-tested, the top score and the category page URL. Not for one product's score or verdict: get_product_score explains a specific product.
list_categories
ChatGPTList every product category Scout ranks, grouped by section (Food & Pantry, Beverages, Nutrition, Home & Personal, Clothing, Family), with the number of hand-reviewed products in each (a hub category shows its sub-categories in brackets; some categories hold no reviewed products yet) and each category's page URL. Answers 'what categories does Scout rank' and 'what does Scout cover'. Takes no input. Not needed before get_category_rankings, which accepts everyday category names and itself lists the closest categories when a name is not recognised.
search_products
ChatGPTSearch Scout's product database by product name, brand or keywords, for example 'Oatly oat milk', 'Liquid Death' or 'Siete tortilla chips'. Answers 'does Scout have X', 'which X do you mean' and 'show me Scout's scores for a brand's range', and identifies the exact product or barcode when a name is loose or a brand sells several versions. Returns up to 20 matches, each with the product's Scout score (1–100) and label, brand, whether it is hand-reviewed, lab-tested or a Scout's Pick, its barcode when it is a catalog entry, and its scouthealthapp.com page URL; hand-reviewed matches come first when the query names them closely, the rest are catalog entries ordered by demand, a missing score means Scout has not scored that entry yet, and the same product can appear more than once under different barcodes or pack sizes. Returns no verdict, ingredients or lab results (get_product_score does) and does not rank a category (get_category_rankings does). Covers packaged food, drinks, supplements and personal care sold in the US; not fresh produce, restaurants or menu items.
search_restaurants
ChatGPTSearch Scout-rated restaurant locations by name, for example 'sweetgreen', 'Cava', 'True Food Kitchen'. Answers 'does Scout rate X', 'is X healthy' for a restaurant without a scored menu, and 'what tier is X' for a named place. Returns up to 10 matching US locations with name, cuisine, address, Scout's tier (Scout's Pick, Excellent or Okay; lower tiers are not published) with its one-line meaning, and the location page URL; the tier line is the whole of Scout's reasoning for a location — there is no menu, dish or ingredient analysis. Name search only: no city, distance or 'near me' lookup, and no results for a restaurant Scout has not rated. A chain's scored menu, best orders and items to skip come from get_chain_menu.