maagpi-youtube-mcp

by vamsi-kodimela

Not rated
GitHub

About

https://github.com/vamsi-kodimela/maagpi-youtube-mcp

Details

Author
vamsi-kodimela
Categories
Other, Media, Search

Setup

Install maagpi-youtube-mcp in your MCP client (Claude Desktop, Cursor, Windsurf, and others).

Repository: https://github.com/vamsi-kodimela/maagpi-youtube-mcp

Follow the installation instructions in the repository README, then restart your MCP client.

maagpi-youtube-mcpis a full YouTube channel-management MCP server: upload videos, schedule publishing, query analytics, moderate comments, manage playlists, update channel branding, and operatemultiple YouTube channels simultaneously— all from any MCP-compatible AI client.

Wraps the official YouTube Data API v3 + YouTube Analytics API behind a single MCP surface with built-in OAuth, quota tracking, response caching, and structured error handling.

- Default:stdio (local process, spawned by your MCP client)
- Optional:Streamable HTTP (YOUTUBE_MCP_TRANSPORT=http) for multi-client / remote access
- Distribution:npx maagpi-youtube-mcp— no global install required
- Full docs:seeTool Referencebelow

This server usesGoogle OAuth 2.0(Desktop app credentials). You provide a Client ID + Client Secret as environment variables; the server handles the browser consent flow and stores refresh tokens to your OS config directory under namedprofiles(one per channel).
-
Google Cloud Console→APIs & Services → Library→ enable:

- YouTube Data API v3
- YouTube Analytics API

Keep your Client Secret out of source control. The server never transmits it anywhere exceptaccounts.google.com.

First call:the server opens a browser window for OAuth consent. Approve once, and tokens are saved automatically under the"default"profile and auto-refreshed thereafter. Add more channels withyoutube_account_add.

Pick your client and drop the snippet into the corresponding config file. Replaceyour_client_id/your_client_secretwith your Google OAuth credentials.

~/Library/Application Support/Claude/claude_desktop_config.json(macOS)%APPDATA%\Claude\claude_desktop_config.json(Windows)

{ "mcpServers": { "youtube": { "command": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } }
{ "mcpServers": { "youtube": { "command": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } }
claude mcp add youtube -- npx -y maagpi-youtube-mcp \ -e YOUTUBE_CLIENT_ID=your_client_id \ -e YOUTUBE_CLIENT_SECRET=your_client_secret
{ "mcpServers": { "youtube": { "command": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } }
{ "mcpServers": { "youtube": { "command": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } }

.vscode/mcp.json(workspace) or User Settings:

{ "servers": { "youtube": { "type": "stdio", "command": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } }
{ "context_servers": { "youtube": { "command": { "path": "npx", "args": ["-y", "maagpi-youtube-mcp"], "env": { "YOUTUBE_CLIENT_ID": "your_client_id", "YOUTUBE_CLIENT_SECRET": "your_client_secret" } } } } }
mcporter add youtube --command "npx -y maagpi-youtube-mcp" \ --env YOUTUBE_CLIENT_ID=your_client_id \ --env YOUTUBE_CLIENT_SECRET=your_client_secret mcporter call youtube youtube_account_list '{}'
YOUTUBE_MCP_TRANSPORT=http \ YOUTUBE_MCP_HTTP_PORT=3000 \ YOUTUBE_CLIENT_ID=your_client_id \ YOUTUBE_CLIENT_SECRET=your_client_secret \ npx -y maagpi-youtube-mcp # MCP endpoint: POST http://127.0.0.1:3000/mcp # Health check: GET http://127.0.0.1:3000/health

Then point any HTTP-capable MCP client athttp://127.0.0.1:3000/mcp.

All tools accept an optionalchannelparameter (profile name). Omit it to use the active profile.

Once connected, send these to your AI client as natural language:

Upload /videos/tutorial.mp4 with title "Getting Started with TypeScript", description "A beginner's guide", tags ["typescript","programming"], unlisted.
Schedule video dQw4w9WgXcQ to go public on January 20 2026 at 3pm UTC.
Get top 5 videos by views in Q1 2025 for my "main" channel, and also for my "gaming" channel.
Show me revenue breakdown for 2025-01-01 to 2025-01-31, split by day.
Reply to comment Ugxxxxx with "Thanks for the feedback! Fixed in v2."

Tools that delete data requireconfirm: true— your AI client will ask before proceeding.

Connect any number of YouTube accounts and target any of them from any tool via the optionalchannelparameter — no switching required.

Add a new YouTube channel profile named "gaming" List all my connected YouTube channel profiles Switch my active YouTube profile to "gaming" Upload /videos/clip.mp4 to my "gaming" channel, title "EP1", public

CLI auth for a named profile (handy for headless setups):

YouTube Data API v3 grants10,000 units/dayby default. Every tool response includes aquotafield:

{ "quota": { "used": 151, "budget": 9000, "remaining": 8849, "resetAt": "2026-01-15T08:00:00.000Z", "warningLevel": "ok", "costOfThisCall": 1 } }

warningLevel:"ok""warn"at 80% →"critical"at 95%.

- GET responses are cached (videos 60s, channels 5min, analytics 5min) — repeat reads cost 0 quota
- Writes auto-retry on 429/5xx with exponential backoff (up to 3×)
- SetYOUTUBE_MCP_QUOTA_LIMITbelow 10,000 to leave headroom

All errors are returned as structured tool content — agents can read and act on them without crashing:

{ "success": false, "error": { "code": "QUOTA_EXCEEDED", "message": "YouTube API daily quota has been exceeded.", "suggestedFix": "Wait for quota reset at midnight Pacific Time, or increase your quota in Google Cloud Console.", "retryable": false, "docsUrl": "https://developers.google.com/youtube/v3/getting-started#quota" }, "quota": { "used": 9001, "warningLevel": "critical" } }
npm install npm run dev # run with tsx (requires .env) npm run auth # authenticate the default channel profile npm run auth -- --channel gaming # authenticate a named profile npm run typecheck # tsc --noEmit npm test # vitest unit tests npm run build # production bundle → dist/index.js npm pack --dry-run # verify publish artifact

- Operate YouTube from any AI client— Claude, Cursor, Windsurf, Zed, VS Code, mcporter
- Manage multiple channels at once— no context switching, target any channel per call
- Production-grade plumbing— OAuth, refresh tokens, quota tracking, retries, structured errors
- Full surface coverage— videos, scheduling, analytics, comments, playlists, branding

- npm:https://www.npmjs.com/package/maagpi-youtube-mcp
- YouTube Data API:
https://developers.google.com/youtube/v3
- YouTube Analytics API:
https://developers.google.com/youtube/analytics
- Model Context Protocol:
https://modelcontextprotocol.io

Find anything in any video. Semantic video search, video Q&A, persistent memory, and social media import (TikTok/YouTube/Instagram) for AI agents. 18 MCP tools.

High-efficiency YouTube MCP server providing token-optimized, structured data for LLMs.

Automate end-to-end youtube production pipeline for face less channels

An MCP server for programmatic control of smartscreen.tv displays via HTTP and MCP commands, with YouTube integration.

No local install, no cloud IP blocks. YouTube transcript, search, and channel data for AI agents. TranscriptAPI runs the infrastructure — 500K+ transcripts/day at 49ms. OAuth or API key. 100 free credits.

Local MCP server for YouTube metadata, thumbnails, comments, and analytics.

MCP server for Al Jazeera 360 — Connect AI assistants to Al Jazeera's streaming catalog. Search, browse, and retrieve Arabic video content with direct watch links.

BYOK Reddit/YouTube research CLI with a tamper-evident audit log and an MCP server.

Interact with the Bilibili video website, enabling actions like searching for videos, retrieving video information, and accessing user data.

An agent-based tool for web search and advanced research, including analysis of PDFs, documents, images, and YouTube transcripts.

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.