MCP Embedding Search

Example Use Cases

• 1

  Retrieve the most relevant transcript segments for a user's question using precomputed embeddings.

• 2

  Precise context with metadata (episode title and timestamps) to cite or display source material alongside answers.

• 3

  Filter or limit results by relevance score and result count to control response size and quality.

Description

This MCP tool accepts a question and returns the most semantically relevant transcript segments using precomputed embeddings stored in a Turso database. Results include episode title, segment text, start/end timestamps, and a similarity score, with configurable result limits and minimum score thresholds. The server is optimized with connection pooling and robust error handling to enable fast, reliable context retrieval without generating new embeddings.