Supabase & PostgreSQL Expert Skill

This skill provides patterns for implementing RAG logic with Supabase and the pgvector extension.

📂 Storage Pattern (Postgres)

• Tables:
  • documents: Stores source document text and global metadata.
  • chunks: Stores text fragments, embedding (vector), and a foreign key document_id (UUID).
• Relationships: Ensure a foreign key relationship with ON DELETE CASCADE from chunks to documents for clean deletions.

🔍 Search Patterns

• Semantic Search (pgvector):

  • Assumes a stored procedure match_chunks exists in the database.
  • Parameters: query_embedding (vector), match_threshold (float), match_count (int).
  • Use self.client.rpc("match_chunks", rpc_params).execute().