I'm building an app in Lovable that relies on web scraping (Zyte API) and AI (OpenAI Responses API) integration. I'm having trouble with some of the finer points of continuation flows and tool calling in OpenAI Responses API.

Hi! I'm building Rowan, a sustainable shopping app built in Lovable. I need an engineer who’s shipped the OpenAI Responses API with function tools + web_search and Supabase continuation (/responses/{id}/tool_outputs). I have a Deno edge function that’s close; it needs schema-correct tool definitions, argument normalization, and robust continuation. If you’ve done this before, could you share 1–2 code snippets or a repo showing tool calls + continuation + structured outputs? Thanks!