MCP Video Digest

MCP Video Digest is an MCP server that downloads and transcribes audio from online video links across over 1,000 supported platforms, including YouTube, Bilibili, TikTok, and Twitter. Designed for researchers, content creators, and developers interacting with AI assistants, it enables language models to extract full transcripts and summarize video content without manually downloading or uploading media files. The server connects directly to external speech-to-text APIs, supporting Deepgram, Gladia, Speechmatics, and AssemblyAI. It dynamically selects the transcription provider based on configured API keys and includes features such as speaker diarization and asynchronous audio processing. Temporary download files are automatically managed and removed after transcription is finished. By executing video extraction pipelines via a unified tool interface, MCP Video Digest allows LLM agents to inspect online multimedia resources, parse spoken dialogues, and generate comprehensive digests directly within an agentic workflow.

Category: Design, Media & Creative

Tags: media, summarization, transcription, video

Visit MCP Video Digest

How to install and configure MCP Video Digest

1. Clone the repository and navigate into the folder: bash git clone https://github.com/R-lz/mcp-video-digest.git cd mcp-video-digest 2. Create and activate a virtual environment using uv: bash uv venv source .venv/bin/activate 3. Install dependencies: bash uv pip install -e . 4. Copy .env.example to .env and configure at least one API key: bash mv .env.example .env Set DEEPGRAM_API_KEY, GLADIA_API_KEY, SPEECHMATICS_API_KEY, or ASSEMBLYAI_API_KEY. 5. Start the server via uv run src/main.py. 6. Connect your MCP client using SSE configuration: json { "mcpServers": { "video_digest": { "url": "http://<ip>:8000/sse" } } }

What you can do with MCP Video Digest

• Transcribing online video content from YouTube, Bilibili, or TikTok directly into an LLM client via a URL. * Generating detailed meeting summaries or podcast lecture notes using automatic speech-to-text and speaker diarization. * Extracting dialogue transcripts from web streaming links across over 1,000 platforms supported by yt-dlp. * Integrating automated video content analysis into multi-agent research workflows using flexible transcription providers like Deepgram or AssemblyAI.

Key facts

• https://github.com/R-lz/mcp-video-digest
• Design, Media & Creative, Web Search & Research
• media, summarization, transcription, video

Part of MCP Servers

What is MCP Video Digest?

MCP Video Digest is a Model Context Protocol server that extracts audio from video links across more than 1,000 streaming websites and transcribes the speech into text using external transcription APIs like Deepgram, Gladia, Speechmatics, or AssemblyAI.

Which transcription services are supported by MCP Video Digest?

The server supports Deepgram, Gladia, Speechmatics, and AssemblyAI. You must configure at least one service API key in your environment file, and the server prioritizes services based on available credentials.

What license does MCP Video Digest use?

MCP Video Digest is open source software released under the MIT License, allowing users to modify, distribute, and integrate the tool into commercial and private projects.

How do I handle authentication for YouTube downloads?

For YouTube downloads that require account authentication or verification, you can export your browser cookies to a cookies.txt file in the project root directory or configure browser cookie extraction supported by yt-dlp.