Can I vibe code Searchspring?

You are building a custom site search and visual merchandising service to replace Searchspring for a multi-platform e-commerce backend.

CORE ARCHITECTURE:
1. Search Engine Core: Deploy a hosted Typesense instance as the primary search and auto-complete engine.
2. Indexing Engine:
   - Construct webhook ingestion endpoints for Shopify (products/create, products/update, products/delete, inventory_levels/connect) and BigCommerce.
   - Design a catalog sync pipeline that transforms raw store products into flattened JSON documents containing: id, title, handle, description, price, compare_at_price, tags, vendor, options, variants, inventory_quantity, total_sales, created_at, and dynamic attributes for faceting.
   - Implement differential sync using product updated_at timestamps to handle delta updates every 15 minutes, with a full night-reindex job.

3. Visual Merchandising Rules Engine:
   - Create a Postgres schema storing: MerchandisingRule (id, query_pattern, match_type [exact|contains], pinned_product_ids [], buried_product_ids [], boosted_tags [], status, start_date, end_date).
   - Middleware layer: Intercept incoming search API requests. Before querying Typesense, evaluate active MerchandisingRules. Inject manual overrides into the Typesense query parameter string (using override_tags, pinned_hits, or custom filter_by syntax).

4. Front-End Autocomplete & Search JS SDK:
   - Build a lightweight standalone Vanilla JS snippet (<12kB) to embed on client storefronts.
   - Render instant search-as-you-type UI dropdown showing products, suggested query terms, and category collections.
   - Support faceted collection filters (price range, color, size, brand, availability) dynamically populated from search response facets.

5. Analytics & Redirect Engine:
   - Track 'zero_results_query' events and store them in Postgres for merchant dashboard reporting.
   - Support Keyword Redirects: If a user searches for a specific rule query (e.g., "shipping status"), return an immediate HTTP 302 redirect payload instead of search results.

OUT OF SCOPE:
- AI conversational chat agents.
- Social feed management and Google Shopping feed generation.
- Personalization based on historical cross-site cookie tracking.

FAILURE MODES TO HANDLE:
- Webhook floods during massive bulk product price updates must be buffered via Redis/BullMQ to prevent hammering the Typesense write lock.
- Out-of-stock SKUs must be automatically demoted to the bottom of the grid unless explicitly pinned by a merchant rule.

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