yt-dlp-mcp

MCP.Pizza Chef: kevinwatt

Ask for the transcript of a talk and it arrives as plain text you can summarize on the spot; ask for the video or just the sound and the file lands in your Downloads folder. It also searches YouTube by keyword and reads back titles, channels, lengths and upload dates. Nothing costs money and no key is required, but the free yt-dlp program has to be installed on your machine first, and your assistant must be running on that same computer.

Files/PDF
Web/Research

Use This MCP server To

Summarize a two-hour talk from its transcript instead of watching it Save a podcast episode as audio to listen to offline Search YouTube for recent videos on a topic I'm researching Pull the subtitles of a lecture in another language Grab one clip from a long video by start and end time Check a video's upload date, channel and length before sharing it

README

🎬 yt-dlp-mcp

A powerful MCP server that brings video platform capabilities to your AI agents

npm version License: MIT Node.js Version TypeScript

Integrate yt-dlp with Claude, Dive, and other MCP-compatible AI systems. Download videos, extract metadata, get transcripts, and more — all through natural language.

Features • Installation • Tools • Usage • Documentation


✨ Features

yt-dlp-mcp FAQ

Can I use this to summarize a long video without watching it?
Yes — ask for the transcript and then for a summary, and the whole thing happens in one conversation.
Do I need an account or a paid key?
No. It is free, and it talks to video sites directly rather than through a paid service.
Is there anything I have to install first?
Yes. The free yt-dlp program must be on your computer — the project lists a one-line install for Windows, Mac and Linux.
Where do the downloads end up?
In your normal Downloads folder, as ordinary video or audio files.
Does this work in the Claude website or a phone app?
No. Everything runs on your own machine, so it only works in a desktop app or editor that reads a local setup file.
Does it only handle YouTube?
Downloads and transcripts work across the many sites yt-dlp supports, including Facebook. Keyword search is YouTube only.
Is it actively maintained?
Yes — the project last shipped code in May 2026.