YouTube Transcript Knowledge Base

Example Use Cases

• 1

  Search across and retrieve relevant segments from YouTube video transcripts using semantic similarity.

• 2

  Answer natural language questions grounded in specific videos with timestamps and source metadata.

• 3

  Organize videos into lists, attach summaries or notes, and filter content by channel, tags, or view counts.

Description

This project captures and processes YouTube video transcripts and metadata, segments them semantically, and builds a FAISS vector index for fast semantic search. It supports multiple embedding backends (OpenAI, Ollama, HuggingFace) with automatic fallbacks, and provides tools to query videos, create lists, add summaries, and jump to timestamps. The MCP server can be connected to compatible clients (for example Claude Desktop) so agents can ask natural language questions and receive context-rich answers referencing specific videos and timestamps.