Can I vibe code Freshrelevance?

Build an open-source alternative to Freshrelevance for an e-commerce platform using Next.js, Cloudflare Workers, Supabase (with pgvector), and Redis.

1. DATA MODEL:
- Track 'events': user_id, session_id, event_type (view_product, add_to_cart, purchase), product_id, timestamp, metadata.
- Store 'product_embeddings': product_id, embedding_vector (156 dims), updated_at.
- Store 'user_profiles': user_id, email, viewed_categories, last_seen, cart_state (jsonb).

2. TRACKING SCRIPT & EDGE API:
- Lightweight lightweight Vanilla JS tracking snippet (< 5kb) to embed on client storefronts.
- Cloudflare Worker endpoint `/api/track` receiving POST requests of user actions. Writes to a Redis Queue for async processing into Supabase.
- Identity Resolution: Map browser cookie `anon_id` to known `user_id` upon email click URL parameter ingestion (`?utm_uid=...`).

3. RECOMMENDATION ENGINE:
- Next.js API route `/api/recommendations` accepting `user_id` or `product_id` and returning:
  a) 'Frequently Bought Together': Computed via Postgres SQL co-occurrence query on completed orders.
  b) 'Recommended For You': Vector cosine similarity query via pgvector against user recent views.

4. DYNAMIC EMAIL IMAGE ENDPOINT:
- Next.js API route `/api/email/dynamic-image.png` that accepts `user_id`.
- Uses Node `@napi-rs/canvas` to generate a 600x300 PNG image on the fly displaying 3 recommended products with titles, prices, and images.
- Sets explicit HTTP headers: `Cache-Control: no-cache, no-store, must-revalidate, max-age=0`, `Pragma: no-cache`, `Expires: 0`.

5. FAILURE MODES:
- If vector processing times out, fallback to top-selling products query.
- If image canvas rendering fails, return a fallback static promotional banner image buffer.

$ each button prefixes agent-specific run instructions · build your own product, never copy proprietary code, trademarks or designs