youtube-personal-feed
MCP server that connects to your personal YouTube account and exposes your subscription feed, letting you list subscribed channels, recent uploads, and get YouTube watch links for handoff to other MCPs.
README
youtube-personal-feed
An MCP server that connects to your personal YouTube account and exposes your subscription feed to AI assistants: list your subscribed channels, list the most recent uploads across your feed, and get https://www.youtube.com/watch?v= links that can be handed off to other MCPs (for example transcript-extraction tools).
Requirements
- Node.js 18+
- A Google account and a Google Cloud project with the YouTube Data API v3 enabled
- An MCP client (opencode, Claude Desktop, etc.)
Google Cloud Console setup (one-time)
-
Go to https://console.cloud.google.com and create a project (or reuse one).
-
Enable the API: APIs & Services -> Library -> search "YouTube Data API v3" -> Enable.
-
OAuth consent screen: APIs & Services -> OAuth consent screen.
- User type: External, then add your own Gmail address as a test user.
- Add the scope
https://www.googleapis.com/auth/youtube.readonly.
-
Create a Desktop client: APIs & Services -> Credentials -> Create credentials -> OAuth client ID -> Application type Desktop app.
- No redirect URI needs to be registered; loopback redirect is allowed for Desktop clients.
-
Copy the Client ID and Client secret into a
.envfile at the repo root:CLIENT_ID=xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxxxxxx.apps.googleusercontent.com CLIENT_SECRET=GOCSPX-xxxxxxxxxxxxxxxxxxxxxxxx(
cp .env.example .envfirst.)
Install and authenticate
Global installation
Install the published package globally to run the CLI and MCP server from any directory:
npm install -g youtube-personal-feed
Create ~/.config/youtube-mcp/.env with your Google Desktop OAuth credentials:
mkdir -p ~/.config/youtube-mcp
cat > ~/.config/youtube-mcp/.env <<'EOF'
CLIENT_ID=xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxxxxxx.apps.googleusercontent.com
CLIENT_SECRET=GOCSPX-xxxxxxxxxxxxxxxxxxxxxxxx
EOF
chmod 600 ~/.config/youtube-mcp/.env
Authenticate once, then run either executable from any directory:
youtube-personal-feed-auth
youtube-personal-feed subscriptions --limit 20
youtube-personal-feed-mcp
The OAuth token is stored at ~/.config/youtube-mcp/token.json. You can instead provide CLIENT_ID and CLIENT_SECRET as environment variables.
Local development
npm install
npm run auth
npm run auth prints a URL, opens your browser, and asks you to consent. Tokens are stored in ~/.config/youtube-mcp/token.json (read-only to you) and are refreshed automatically at runtime.
Note: because this uses an unverified "Desktop app" client, Google shows a "Google hasn't verified this app" warning the first time. Click Advanced -> Go to <project> (unsafe) to proceed.
Register with an MCP client
-
opencode: a project-level
.mcp.jsonis included. It registers the server asyoutube-personal-feedwithnpx tsx src/index.ts. Reopen opencode in this directory to pick it up. -
Claude Desktop: add to
claude_desktop_config.json:{ "mcpServers": { "youtube-personal-feed": { "command": "youtube-personal-feed-mcp" } } } -
Local development: use
npx tsx src/index.tswithcwdset to the repository.
Tools
| Tool | Inputs | Output |
|---|---|---|
list_subscriptions |
query?, maxResults? (default 50) |
channelId, title, channelUrl, thumbnail, publishedAt |
list_feed |
limit? (default 20), sinceDays?, channelId? |
latest uploads across subscriptions: videoId, title, channelTitle, publishedAt, videoUrl |
list_channel_uploads |
channelId, maxResults? (default 15) |
uploads for any channel, same fields |
get_video |
videoId |
snippet/contentDetails/statistics + videoUrl + channelUrl |
Feed items always include videoUrl in the form https://www.youtube.com/watch?v=<id> — the format accepted by most transcript-extraction MCP servers.
list_feed vs. list_channel_uploads
| Use case | Tool | Behavior |
|---|---|---|
| See recent videos from all subscriptions | list_feed |
Fetches recent uploads across your subscriptions and sorts the combined results by publication date. |
| See recent videos from one subscription | list_feed with channelId |
Limits the subscription feed to that channel. |
| See recent videos from any known channel | list_channel_uploads |
Fetches that channel directly; it does not need to be one of your subscriptions. |
For example, first call list_subscriptions to obtain a subscribed channel's channelId, then pass it to list_feed:
{
"channelId": "UCJaGVXG4KgOUXUtcmAHAOdA",
"limit": 10
}
Use list_channel_uploads with that same ID if the channel is not in your subscriptions, or if you specifically want a direct per-channel lookup.
CLI
The CLI uses the same OAuth credentials and token as the MCP server. Run youtube-personal-feed after global installation, or use npm run cli -- while developing:
youtube-personal-feed subscriptions --query running --limit 20
youtube-personal-feed feed --limit 10 --since-days 7 # all subscriptions
youtube-personal-feed feed --channel-id UCJaGVXG4KgOUXUtcmAHAOdA # one subscription
youtube-personal-feed uploads UCJaGVXG4KgOUXUtcmAHAOdA --limit 15 # any channel
youtube-personal-feed video 64wtzsSQx84
For local development:
npm run cli -- subscriptions --query running --limit 20
npm run cli -- feed --limit 10 --since-days 7
npm run cli -- feed --channel-id UCJaGVXG4KgOUXUtcmAHAOdA --limit 10
npm run cli -- uploads UCJaGVXG4KgOUXUtcmAHAOdA --limit 15
npm run cli -- video 64wtzsSQx84
Each command writes JSON to standard output. Run youtube-personal-feed --help (or npm run cli -- --help) for the command reference.
Quota
The YouTube Data API default quota is 10,000 units/day. List calls cost ~1 unit each. This server caches subscriptions for 30 minutes and per-channel uploads for 10 minutes, so a full feed refresh over ~100 subscriptions costs about 100 units. If you hit the quota, the tools return a clear error; wait until the next day or request more quota.
Development
npm run typecheck # type-check
npm run lint # lint and check formatting
npm run format # apply source formatting
npm run build # compile to dist/
npm run smoke # sanity check auth + subscriptions (prints real data)
npm install installs a pre-commit hook. Commits run staged TypeScript files through Biome and then run npm run typecheck. Bypass this only when necessary with git commit --no-verify.
Project layout
src/config.ts # env loading, credential + token path helpers
src/auth.ts # one-time OAuth loopback flow (npm run auth)
src/youtube.ts # YouTube Data API v3 wrapper + caching
src/index.ts # MCP server (McpServer, stdio, 4 tools)
src/types.ts # shared types
scripts/smoke.ts
Recommended Servers
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.