YouTube MCP
Provides production-grade tools for YouTube channel resolution, video metadata extraction, transcripts, and playlist management. It features a quota-aware, AI-friendly design that supports structured searching and listing of public YouTube data.
README
YouTube MCP (youtube_mcp)
A Model Context Protocol (MCP) server that provides production-grade YouTube inventory tools (channel resolution, channel videos, channel playlists, playlist videos) designed to be safe-by-default, quota-aware, and AI-friendly.
Repository: https://github.com/glonorce/youtube_mcp PyPI: https://pypi.org/project/glonorce-youtube-mcp/
Naming note:
- Repo name:
youtube_mcp- Python package / import name:
youtube_mcp- MCP entry module:
python -m youtube_mcp.server
Prerequisites (PATH)
Make sure these commands work in your terminal (they must be on PATH):
python --versionpython -m pip --versionpipx --version(only if you use the pipx-based MCP config)
<details> <summary>✅ What is <code>PATH</code>?</summary>
PATH is the list of folders that your OS searches when you run a command like python or pipx.
- If a command is not on PATH, you'll see errors like:
- Windows:
executable file not found in %PATH% - PowerShell:
... is not recognized as the name of a cmdlet
- Windows:
Quick checks (Windows):
where python
where pipx
$env:Path
</details>
On Windows:
- install Python from https://www.python.org/downloads/
- during installation, check "Add Python to PATH"
Install pipx (only if you use the pipx-based MCP config):
python -m pip install --user pipx
python -m pipx ensurepath
Notes:
- After
pipx ensurepath, restart your terminal/IDE sopipxis available onPATH. - Verify with:
pipx --version
✨ Available on PyPI
pip install glonorce-youtube-mcp
# run
youtube_mcp
Screenshots
<!-- GitHub README does not allow JavaScript, so we can't implement a true clickable carousel. This is a "scroll-snap" horizontal gallery (swipe/trackpad/shift+wheel to move sideways). -->
<div style="display:flex; gap:12px; overflow-x:auto; padding: 8px 0; scroll-snap-type:x mandatory; -webkit-overflow-scrolling: touch;"> <a id="screenshot-test-plan" style="scroll-snap-align:start; flex:0 0 auto;" href="#screenshot-test-plan"> <img src="screenshots/test_plan.png" alt="Test plan" style="height: 360px; display:block;" /> </a> <a id="screenshot-test-execution" style="scroll-snap-align:start; flex:0 0 auto;" href="#screenshot-test-execution"> <img src="screenshots/test_execution.png" alt="Test execution" style="height: 360px; display:block;" /> </a> <a id="screenshot-success-result" style="scroll-snap-align:start; flex:0 0 auto;" href="#screenshot-success-result"> <img src="screenshots/success_result.png" alt="Success result" style="height: 360px; display:block;" /> </a> </div>
Features
- 7 MCP tools (1 text-output tool + 6 structured-output tools)
- Safe defaults (public data only; shorts/live excluded unless enabled)
- Quota-aware design (budgeting + expensive strategies as explicit opt-in)
- Tool discovery works in MCP hosts (tools registered with rich docstrings)
- Resilient YouTube Data API client (endpoint allowlist, retries, gzip handling)
Tools
This server exposes 7 tools:
-
get_yt_video_info(video_id_or_url) -> str- Accepts a raw video id or a full YouTube URL.
- Returns video metadata + transcript (when available).
-
resolve_youtube_channel(channel_ref, ...) -> dict -
list_youtube_channel_videos(channel_ref, ...) -> dict- Paginated.
- Returns
nextPageTokenandnext_page_token(alias).
-
list_youtube_channel_playlists(channel_ref, ...) -> dict- Paginated.
- Returns
nextPageTokenandnext_page_token(alias).
-
list_youtube_playlist_videos(playlist_id, ...) -> dict- Paginated.
- Returns
nextPageTokenandnext_page_token(alias).
-
search_youtube_channel_videos(channel_ref, query, ...) -> dict- Keyword search within a channel (quota-expensive, uses Search API).
- Paginated.
-
list_youtube_video_comments(video_id, ...) -> dict- Public comment threads for a video (comments may be disabled).
- Paginated.
All tools except get_yt_video_info return structured JSON and are documented with AI-friendly docstrings.
Installation & MCP configuration
PyPI distribution name:
glonorce-youtube-mcpRepo name:
youtube_mcp· import name:youtube_mcp
API key
All inventory/search/comment tools require a YouTube Data API v3 key:
YOUTUBE_API_KEY
Option 1 (recommended for MCP hosts): pipx (no repo clone)
Use this when you want a "zero local project" setup (no clone, no venv management).
{
"mcpServers": {
"youtube_mcp": {
"command": "pipx",
"args": ["run", "--spec", "glonorce-youtube-mcp", "youtube_mcp"],
"env": {
"YOUTUBE_API_KEY": "YOUR_KEY"
}
}
}
}
Option 2: pip into a dedicated venv (no repo clone)
python -m venv .venv_youtube_mcp
# Windows: .venv_youtube_mcp\Scripts\activate
# macOS/Linux: source .venv_youtube_mcp/bin/activate
python -m pip install -U pip
python -m pip install glonorce-youtube-mcp
MCP host JSON example (simple):
{
"mcpServers": {
"youtube_mcp": {
"command": "youtube_mcp",
"env": {
"YOUTUBE_API_KEY": "YOUR_KEY"
}
}
}
}
If your MCP host can't find youtube_mcp on PATH (Windows error like: executable file not found in %PATH%), use an absolute path.
MCP host JSON example (absolute path to venv Python + module):
{
"mcpServers": {
"youtube_mcp": {
"command": "C:\\PATH\\.venv_youtube_mcp\\Scripts\\python.exe",
"args": ["-m", "youtube_mcp.server"],
"env": {
"YOUTUBE_API_KEY": "YOUR_KEY"
}
}
}
}
Option 3: local clone (development)
git clone https://github.com/glonorce/youtube_mcp.git
cd youtube_mcp
python -m venv .venv
# Windows: .venv\Scripts\activate
# macOS/Linux: source .venv/bin/activate
python -m pip install -e .
MCP host JSON example (local clone):
{
"mcpServers": {
"youtube_mcp": {
"command": "C:\\PATH\\youtube_mcp\\.venv\\Scripts\\python.exe",
"args": ["-m", "youtube_mcp.server"],
"env": {
"YOUTUBE_API_KEY": "YOUR_KEY"
}
}
}
}
Getting Your YouTube API Key
While not required, you can optionally set up a YouTube Data API key for enhanced functionality. Here's how to get one:
Step 1: Create a Google Cloud Project
- Go to the Google Cloud Console
- Click "Select a project" at the top of the page
- Click "New Project" and give it a name (e.g., "MCP YouTube Extract")
- Click "Create"
Step 2: Enable the YouTube Data API
- In your new project, go to the API Library
- Search for "YouTube Data API v3"
- Click on it and then click "Enable"
Step 3: Create API Credentials
- Go to the Credentials page
- Click "Create Credentials" and select "API Key"
- Your new API key will be displayed - copy it immediately
- Click "Restrict Key" to secure it (recommended)
Step 4: Restrict Your API Key (Recommended)
- In the API key settings, click "Restrict Key"
- Under "API restrictions", select "Restrict key"
- Choose "YouTube Data API v3" from the dropdown
- Click "Save"
Step 5: Set Up Billing (May be required)
- Go to the Billing page
- Link a billing account to your project
- Note: YouTube Data API has a free tier of ~10,000 units per day, which is typically sufficient for most use cases
API Key Usage Limits
- Free Tier: ~10,000 units per day (commonly)
- Cost: depends on your Google Cloud / billing configuration
Security Best Practices
- Never commit your API key to version control
- Use environment variables as shown in the MCP configuration section
- Restrict your API key to only the YouTube Data API
- Monitor usage in the Google Cloud Console
Example MCP calls (JSON payloads)
Exact envelope depends on your MCP client; these show
tool+arguments.
Resolve a channel
{
"tool": "resolve_youtube_channel",
"arguments": {
"channel_ref": "@GoogleDevelopers",
"resolution_mode": "strict",
"include_uploads_playlist": true
}
}
List channel videos (safe default)
{
"tool": "list_youtube_channel_videos",
"arguments": {
"channel_ref": "@GoogleDevelopers",
"max_videos": 50,
"page_token": null,
"include_shorts": false,
"include_live": false,
"parts_level": "basic",
"order_strategy": "uploads_playlist",
"order_by": "date"
}
}
Quota reference (best-effort)
This section is a best-effort summary and may change over time. Always verify with official documentation.
Last verified: 2026-02-18
Official references:
- YouTube Data API quota calculator / costs: https://developers.google.com/youtube/v3/determine_quota_cost
- Usage limits: https://developers.google.com/youtube/v3/getting-started#quota
Approximate quota usage by tool
Note: get_yt_video_info returns text (not structured JSON) and is not included in the quota table because its behavior depends on optional extractors.
| Tool | Primary endpoints | Typical cost (rough) | Notes |
|---|---|---|---|
resolve_youtube_channel |
channels.list |
~1 | strict mode mostly channels.list; best_effort may use search.list |
list_youtube_channel_videos (default) |
channels.list + playlistItems.list + videos.list |
~ (1 + 1 + 1) per page | uploads-playlist strategy |
list_youtube_channel_videos (order_strategy=search_api) |
search.list + videos.list |
~ (100 + 1) per page | expensive; capped behavior |
list_youtube_channel_playlists |
playlists.list |
~1 per page | public playlists only |
list_youtube_playlist_videos |
playlistItems.list + videos.list |
~ (1 + 1) per page | public playlist |
search_youtube_channel_videos |
search.list + videos.list |
~ (100 + 1) per page | keyword search within a channel |
list_youtube_video_comments |
commentThreads.list |
~1 per page | comments may be disabled (403) |
Daily quota:
- Common free quota: 10,000 units/day
- Reset: daily (Google-defined; see official docs/console)
Testing
Use the venv interpreter to avoid accidentally running global pytest.
# recommended
python -m pytest -q
# if you use a venv
.venv\Scripts\python.exe -m pytest -q
Troubleshooting
Only get_yt_video_info is visible (other tools missing)
Your MCP host is running an old installed copy from site-packages.
Fix:
python -m pip install -e .
Then restart the MCP host.
YOUTUBE_API_KEY placeholder is not expanded in some MCP hosts
Some MCP hosts do not interpolate placeholders like ${YOUTUBE_API_KEY} inside mcp.json.
❌ Not recommended (may be taken literally):
{
"env": { "YOUTUBE_API_KEY": "${YOUTUBE_API_KEY}" }
}
✅ Recommended (explicit value in config):
{
"env": { "YOUTUBE_API_KEY": "YOUR_REAL_KEY_HERE" }
}
✅ Also works (set it in the host process environment, then reference nothing in JSON):
- set
YOUTUBE_API_KEYin the MCP host environment (OS/IDE) - then keep
mcp.jsonwithout placeholders
License
MIT — see LICENSE
Acknowledgements
This project was originally derived from / inspired by https://github.com/sinjab/mcp_youtube_extract. The codebase has since been heavily rewritten and extended.
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.