mcp-youtube

YouTube MCP Server

mcp-youtube brings YouTube transcripts and compact video, channel, and comment data into AI-agent workflows. It serves content creators, marketers, and analysts, with transcript access available without an API key and optional MongoDB caching for API-backed requests.

Server facts

mcp-youtube

Project

Author
kirbah
License
MIT

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
32
Last commit
Data as of

Tags

  • youtube ytcomment subtitles

Overview

YouTube transcript analysis is a central workflow in mcp-youtube: getTranscripts can return complete caption text or focus on opening and closing segments for examining hooks and calls to action. It accepts multiple video IDs and a language option. Beyond transcripts, the project documents filtered video and channel searches, video metadata and engagement ratios, and comment retrieval with controls for sorting, result limits, and replies. These responses are condensed for language-model use rather than exposing the full YouTube API payload.

Channel research includes subscriber, view, and video counts, alongside tools for finding a channel’s strongest-performing videos. Regional trend discovery and category lookup support exploration of YouTube content. For niche research, findConsistentOutlierChannels looks for channels that repeatedly outperform their size; this feature requires MongoDB to retain and analyze channel data. MongoDB also provides optional caching for repeated API requests. Transcript retrieval works without an API key, while search and analytics use a YouTube Data API key. The project runs on Node.js and communicates with MCP clients over stdio.

mcp-youtube setup notes

  • mcp-youtube records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers