ShortsMonkey MCP
A read-only MCP server for discovering YouTube outliers, analyzing viral Shorts, and evaluating single-video performance using stored snapshots.
README
<p align="center"> <img src="assets/icon.png" alt="ShortsMonkey" width="112" height="112" /> </p>
ShortsMonkey MCP
English | 简体中文
ShortsMonkey MCP is a read-only remote Model Context Protocol server for YouTube outlier discovery, viral Shorts rankings, and single-video performance analysis.
It gives MCP-compatible AI assistants access to stored ShortsMonkey data so creators, agencies, researchers, and content teams can find low-subscriber videos with unusually high views and turn those examples into research-backed content decisions.
Connect to the hosted server
Remote Streamable HTTP endpoint:
https://mcp.shortsmonkey.com/mcp
Create a smcp_live_... token at ShortsMonkey MCP Access, then configure a client that supports remote Streamable HTTP and custom headers.
Cursor configuration:
{
"mcpServers": {
"shortsmonkey-mcp": {
"url": "https://mcp.shortsmonkey.com/mcp",
"headers": {
"Authorization": "Bearer smcp_live_your_token_here"
}
}
}
}
Do not commit the token or place it in a shared configuration. See the complete user guide for account, quota, and token-rotation instructions.
Tools
get_hot_today
Returns the current ShortsMonkey hot-today YouTube Shorts leaderboard from stored database snapshots.
| Input | Description |
|---|---|
region |
Optional region code such as US, JP, BR, or IN |
tier |
S, A, B, C, or all; defaults to all |
limit |
1-50 results; defaults to 20 |
Example: Find today's 10 hottest YouTube Shorts in the US.
search_outlier_videos
Searches stored outlier snapshots by keyword, niche, region, format, views, subscribers, and views-to-subscribers ratio.
| Input | Description |
|---|---|
keyword |
Optional title, channel, tag, or source-keyword search |
niche |
Optional niche or category text |
videoType |
shorts, long, or all; defaults to all |
region |
Optional region code |
maxChannelSubscribers |
Maximum channel subscriber count |
minViews |
Minimum video view count |
minVsRatio |
Minimum views/subscribers ratio; defaults to 10 |
timeWindow |
24h, 3d, 7d, or 30d; defaults to 7d |
limit |
1-50 results; defaults to 20 |
Example: Find long-form AI videos from the last 7 days with fewer than 50,000 subscribers and a V/S ratio above 20.
analyze_video_url
Analyzes one YouTube watch, Shorts, embed, or youtu.be URL using a stored ShortsMonkey performance snapshot.
| Input | Description |
|---|---|
url |
Required YouTube video URL |
Example: Analyze this YouTube video and explain whether it is an outlier: <URL>
All three tools declare readOnlyHint: true, destructiveHint: false, idempotentHint: true, and openWorldHint: false.
Returned data
Depending on the tool, responses can include:
- video and channel URLs
- channel subscriber count and video views
- views-to-subscribers ratio and viral score
- Shorts or long-form classification
- niche, region, publish time, and snapshot time
- a concise numeric explanation of why the result is an outlier
Values come from stored ShortsMonkey snapshots and may differ from the current numbers displayed by YouTube.
Quotas and authentication
- Authentication uses
Authorization: Bearer smcp_live_.... - Trial accounts receive five MCP tool calls during the trial.
- Paid accounts receive 200 calls per day, shared with ShortsMonkey website searches.
- Each account can keep up to three active tokens.
- Default HTTP protection limits are 90 requests per token per minute and 180 requests per IP per minute.
Tokens can be created, rotated, and revoked from MCP Access.
Health checks
GET https://mcp.shortsmonkey.com/health
GET https://mcp.shortsmonkey.com/ping
GET https://mcp.shortsmonkey.com/.well-known/mcp/server-card.json
/ping is an alias provided for Glama hosting probes. A browser GET to /mcp returns 405 Method Not Allowed by design because MCP protocol requests use POST.
The public server card exposes only static MCP metadata so registries can discover the three tools without receiving a user token; tool calls still require authentication.
Architecture
MCP client
-> ShortsMonkey MCP gateway (this repository)
-> private ShortsMonkey application API
-> stored YouTube research snapshots
The gateway validates account tokens through the private ShortsMonkey application, applies rate limits, registers MCP tools, and converts internal JSON responses into MCP text and structured content.
The repository intentionally does not contain the ShortsMonkey database, account system, billing logic, internal service credentials, or proprietary data-query implementation.
Development
Requirements:
- Node.js 20 or newer
- npm
- authorized access to the private ShortsMonkey internal API
npm ci
cp .env.example .env
npm run dev
The local endpoint defaults to http://localhost:3001/mcp. Health checks are available at /health and /ping.
Run all checks:
npm run check
Required server environment
| Variable | Purpose |
|---|---|
SHORTSMONKEY_INTERNAL_API_BASE_URL |
Base URL of the ShortsMonkey application |
SHORTSMONKEY_INTERNAL_SERVICE_TOKEN |
Private server-to-server credential |
MCP_ALLOWED_ORIGINS |
Comma-separated browser origins allowed in production |
MCP_IP_RATE_LIMIT_PER_MINUTE |
Per-instance source-IP request limit |
MCP_TOKEN_RATE_LIMIT_PER_MINUTE |
Per-instance authenticated-principal request limit |
MCP_ALLOW_LEGACY_KEYS is disabled by default and must remain disabled in normal production deployments.
Docker
docker build -t shortsmonkey-mcp .
docker run --rm --env-file .env -p 3001:3001 shortsmonkey-mcp
The image runs as the unprivileged node user and includes a /ping health check.
Glama
This repository includes a root glama.json, a production Dockerfile, static tool schemas, annotations, tests, and a complete Glama submission checklist.
Repository URL for submission:
https://github.com/cuilinhao/youtube-mcp
Security
Read SECURITY.md and report vulnerabilities privately to support@shortsmonkey.com.
Source-use notice
This repository currently has no open-source license grant. The source is published for transparency and review. Contact support@shortsmonkey.com before reuse or redistribution outside contributions to this repository.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。