supabase/headless-vector-search

Automates markdown documentation ingestion via GitHub Actions to populate a `docs` schema with vector embeddings, then serves a Supabase Edge Function that converts user queries into streamed ChatGPT-style responses using OpenAI's completions API. The headless architecture decouples search logic from frontend, enabling integration into existing documentation sites through a simple HTTP endpoint that supports server-sent events for streaming responses.

193 stars. No commits in the last 6 months.