DeepSRT YouTube Summarizer

Example Use Cases

• 1

  Fetch and display timestamped YouTube transcripts (cleaned, entity-decoded, and formatted) for user-facing chat or browsing experiences.

• 2

  Generate concise, localized summaries of YouTube videos in narrative or bullet formats to provide quick overviews.

• 3

  Integrate YouTube metadata (title, author, duration) alongside transcripts and translated titles for downstream processing or indexing.

Description

This MCP server integrates directly with YouTube's InnerTube and caption APIs to fetch transcripts and with DeepSRT's worker API to produce summaries and translated titles. It provides two tools: get_transcript (timestamped, cleaned transcripts) and get_summary (narrative or bullet summaries in multiple languages). The server runs on Node/Bun, supports direct CLI usage or stdio MCP server mode, and prioritizes manual captions while gracefully falling back to auto-generated tracks. No pre-caching is required for transcript extraction, though CDN-cached summaries may affect availability.

Capabilities(2 total)