What is the DS.media AI Agent?

It's a Node.js + TypeScript pipeline that ingests YouTube videos, transcribes them using a subtitle-first approach (YouTube subtitles or Whisper ASR), and generates structured JSON analysis (summaries, tags, key takeaways) powered by Ollama.

How does it ensure stable and traceable analysis?

The agent uses the YouTube video ID as the canonical identifier for all generated artifacts (audio, transcript, and JSON analysis), ensuring stable, source-linked filenames and safe batch processing.

Can I integrate DS.media AI Agent with other AI tools?

Yes, it functions as an MCP server, exposing composable tools that allow seamless integration with AI agents like Claude Desktop and Open WebUI, enabling advanced automation and command execution.

Does it support processing multiple YouTube videos at once?

Absolutely. It features robust batch processing capabilities for YouTube playlists, channels, or custom URL lists. It intelligently manages new, partial, and already-processed videos with retry logic.

What kind of output analysis does it provide?

For each video, it produces a validated JSON file containing the video ID, original source URL, a model-generated summary, relevant tags, and concise key takeaways derived from the video's transcript.

DS.media AI Agent: YouTube Video Analysis & JSON Output

The DS.media AI Agent provides a robust Node.js and TypeScript pipeline for processing YouTube videos. It intelligently downloads video audio, prioritizes official YouTube subtitles, or falls back to Whisper transcription when subtitles are unavailable. The extracted transcripts are then summarized using a local Ollama model, generating validated structured JSON analysis, including summaries, tags, and key takeaways, with a consistent naming convention tied to the original video ID. This tool supports both single video ingestion and efficient batch processing of playlists, channels, or custom URL lists, and can be run as a CLI, an MCP server, or integrated with platforms like Claude Desktop and Open WebUI.

主な機能

01Comprehensive YouTube video ingestion pipeline (audio download, subtitle fetching, transcription)

02Ollama-powered summarization producing structured JSON analysis

03Intelligent transcript generation: prioritizes YouTube subtitles, falls back to Whisper ASR

04Stable, source-linked artifact naming using YouTube video IDs for all outputs

05Flexible deployment: CLI, MCP server for AI agent integration (Claude Desktop, Open WebUI), and batch processing capabilities

060 GitHub stars

ユースケース

01Integrating video processing and analysis capabilities into AI agent workflows for advanced conversational AI applications.

02Automated analysis and summarization of YouTube video content for research or content management.

03Batch processing of YouTube playlists or channels to extract structured insights from large video collections.