You provide recipe URLs, free-form shopping list text, and/or an image via content_urls, content_text, content_image, and supermarket domain. Pepesto uses these inputs to build a complete shopping cart.

A session_id representing a fully prepared checkout-driving session. Internally, Pepesto runs the equivalent of /parse, /products, and /session, selecting the best product for each generic item according to Pepesto’s heuristics (user preferences, price, waste minimization, quality, etc.). The returned session_id can be passed directly to /checkout to drive the supermarket cart.

A continue_session_id referencing an existing checkout-driving session, plus optional prev_turn_response (JavaScript execution result) and screenshot payload from the previous turn. Each call advances the same session one step further.

A NextCheckoutIterationResponse containing the current session_id, exactly one instruction to execute (e.g. load_page, await_element, run_js, prompt_user_action, await_js_out_change, or done), and an optional attach_screenshot_on_next_turn flag telling the client to send a screenshot on the next call.

A recipe provided as free text, a publicly crawlable URL, or an image. Optional fields let you specify locale and whether to generate a shareable image.

Structured recipe data (title, ingredients, nutrition, and more) plus a KgToken that you can pass to the /products endpoint to build a fully itemized shopping cart with concrete products, quantities, and images.

Explicit user query and optional preferences and constraints for meal planning, such as ingredients to include or avoid, cuisine, dietary tags, time constraints, and number of servings. The endpoint uses these criteria to search Pepesto’s recipe graph.

A list of suggested recipes from Pepesto’s recipe database (1M+ recipes), in the same format returned by /parse. Each recipe includes rich metadata (title, ingredients, nutrition, instructions, etc.) and a KgToken that can be passed to the /products endpoint to build a fully itemized shopping cart.

One or more recipe KgToken values from the /parse or /suggest endpoints, an optional manual shopping list as free text, and the target supermarket (e.g. domain or ID). The service uses these inputs to build a combined cart for all requested recipes.

A fully itemized shopping cart with concrete products, recent prices, quantities, images, and other metadata. Items are merged across recipes to reduce waste and cost, and for each generic ingredient (e.g. “tomatoes”) multiple matching products are returned so the user or client app can choose the best option.

A list of concrete products selected from the /products response (each wrapped in a ProductAndSuggestedQuantity), along with the target supermarket_domain and user_locale. The list represents the final basket the user intends to purchase.

A shopping session object that encapsulates the user’s final basket for the chosen supermarket, including a session_id that can be passed to the /checkout endpoint, plus summary information about the cart.

A supermarket_domain identifying the retailer whose catalog you want to fetch (e.g. "coop.ch"). Optionally, you can also provide a webhook_url that Pepesto will call whenever the supermarket’s product catalog is re-indexed.

A list of all indexed products for the given supermarket in Pepesto’s database, including names, prices, images, identifiers, and other metadata required to build storefronts, comparison tools, or custom experiences. If a webhook_url is provided, Pepesto will also send incremental updates to that URL each time the catalog is re-indexed.