supabase-community/nextjs-openai-doc-search

Chunks `.mdx` documentation into sections and generates OpenAI embeddings at build time, storing them in Postgres with `pgvector` for efficient semantic search. At runtime, user queries are embedded and matched against stored vectors to retrieve relevant content, which is then injected into GPT-3 completion prompts and streamed to clients via Edge Functions. Includes automated checksum tracking to regenerate embeddings only when source files change, with one-click Vercel deployment and Supabase integration.

1,715 stars. No commits in the last 6 months.