mcp-youtube

mcp-youtube

Enables conversational YouTube playlist management via Claude, allowing creation, pruning, cleanup, and deletion of playlists through the YouTube Data API v3.

Category
Visit Server

README

<div align="center">

TypeScript Node.js npm MIT

▶️ MCP server for the full YouTube Data API v3 — playlists, videos, channels, community, captions, and media, conversationally from any MCP client

<a href="https://kud.io/projects/mcp-youtube">Website</a> · <a href="https://kud.io/projects/mcp-youtube/docs">Documentation</a>

</div>

Features

  • Full YouTube Data API v3 coverage — 52 tools spanning playlists, videos, channels & sections, subscriptions, comments, captions, thumbnails, watermarks, playlist cover images, and read-only reference data, all under the youtube.force-ssl scope
  • Quota-aware by design — every tool documents its cost up front, from 1-unit reads to the 1600-unit upload-video
  • Guarded destructive actions — every delete, unsubscribe, moderation action, or other irreversible/outward tool refuses to run without confirm: true
  • Full-replace done safely — YouTube's update methods overwrite an entire resource part; tools like update-video, update-playlist, update-channel, and update-channel-section read the current resource first so an update that only changes one field never silently blanks the rest
  • Safe bulk cleanupclean-playlist scans for [Deleted video]/[Private video] tombstones, duplicates, and videos from named channels, then defaults to a dry run so you see the plan before anything is removed
  • Local-file media uploads — videos, captions, thumbnails, watermarks, and playlist cover images all upload from a local file path via resumable upload where the API requires it
  • One-time OAuth setup — a desktop-flow npm run setup handles Google authorisation and prints a refresh token; you supply it (and the client id/secret) to the server via environment variables, from any store you like
  • Typed end to end — Zod-validated tool schemas and a Vitest suite covering every handler

Install

The server works with any MCP-compatible client — Claude Desktop, Claude Code, Cursor, Cline, Zed, Continue, and others.

Install the package:

npm install -g @kud/mcp-youtube

In Claude Code or Claude Desktop you can instead install it as a plugin, which wires up the registration for you:

/plugin install youtube@kud

Either way, run the one-time OAuth setup before first use:

npm run setup

This walks you through creating a Google Cloud OAuth client (Desktop app type), opens a browser to authorise the youtube.force-ssl scope, and prints the resulting credentials — it writes nothing to disk. Stash them wherever you keep secrets (keychain, secrets manager, your MCP client's env block) and expose them to the server as environment variables:

Variable Holds
MCP_YOUTUBE_CLIENT_ID OAuth client ID
MCP_YOUTUBE_CLIENT_SECRET OAuth client secret
MCP_YOUTUBE_REFRESH_TOKEN long-lived refresh token

The server reads these three from the environment only — it's store-agnostic, so how they get there is up to you. A keychain-backed export works well:

export MCP_YOUTUBE_REFRESH_TOKEN=$(security find-generic-password -s mcp-youtube-refresh-token -w)

Tip: set the OAuth app's publishing status to In production in Google Cloud. youtube.force-ssl is a sensitive scope, and while the app sits in Testing the refresh token expires after 7 days.

Configuration

Register the server with your MCP client. The exact config file and its location depend on the client — Claude Desktop, for example, uses claude_desktop_config.json (Settings → Developer → Edit Config); Cursor, Cline, Zed, and Continue each have their own. The registration follows the standard mcpServers shape:

{
  "mcpServers": {
    "youtube": {
      "command": "npx",
      "args": ["-y", "@kud/mcp-youtube"],
      "env": {
        "MCP_YOUTUBE_CLIENT_ID": "your-client-id",
        "MCP_YOUTUBE_CLIENT_SECRET": "your-client-secret",
        "MCP_YOUTUBE_REFRESH_TOKEN": "your-refresh-token"
      }
    }
  }
}

If you installed via /plugin install youtube@kud, the plugin handles registration — you only need to supply the three environment variables above.

Usage

YouTube's Data API v3 gives every project a fixed 10,000 quota units/day. A single careless search, bulk delete, or video upload can burn through a meaningful chunk of that, so every tool documents its cost up front — and every destructive or outward-irreversible tool requires an explicit confirm: true.

52 tools cover the full read/write surface reachable under the youtube.force-ssl scope:

Category Tools Docs
Searching search, list-playlists, get-playlist Searching
Playlists create-playlist, update-playlist, add-to-playlist, update-playlist-item Creating & Adding
Cleaning clean-playlist Cleaning Playlists
Deleting remove-from-playlist, delete-playlist Deleting & Pruning
Videos list-videos, update-video, rate-video, get-video-rating, delete-video, upload-video, report-video-abuse Videos
Channels & sections update-channel, list-channels, list-channel-sections, create-channel-section, update-channel-section, delete-channel-section Channels & Sections
Community list-subscriptions, subscribe, unsubscribe, list-comment-threads, create-comment-thread, list-comments, reply-to-comment, update-comment, delete-comment, set-comment-moderation-status Community
Captions & media list-captions, upload-caption, update-caption, download-caption, delete-caption, set-thumbnail, set-watermark, unset-watermark, list-playlist-images, upload-playlist-image, update-playlist-image, delete-playlist-image Captions & Media
Reference data list-activities, list-video-categories, list-i18n-languages, list-i18n-regions, list-video-abuse-report-reasons, list-members, list-membership-levels Reference Data

Once installed, just ask your MCP client:

> Clean up the tombstones in my "Focus" playlist

Scanned 214 items — found 6 tombstones ([Deleted video]/[Private video]).
This is a dry run, no changes made. Estimated cost to delete: 300 units.
Want me to go ahead?

> Yes, and dedupe it too

Deleted 6 tombstones and 3 duplicates (9 items, 450 units).

clean-playlist always returns its plan first — pass dryRun: false (or just confirm in conversation) to actually delete.

Development

git clone https://github.com/kud/mcp-youtube.git
cd mcp-youtube
npm install
npm run dev
Script Purpose
npm run dev Run the server directly from source (tsx)
npm run build Compile to dist/
npm test Run the Vitest suite
npm run typecheck Type-check without emitting
npm run inspect:dev Launch the MCP Inspector against source

📚 Full documentation → mcp-youtube/docs

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured