Ultra MCP SS

★ 1

from motebus

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

🔥🔥🔥✓ VerifiedFreeQuick setup

⌁ labsco summary — our analysis, not the vendor's

What it is — A FastAPI-based MCP server that programmatically controls smartscreen.tv displays over HTTP/MCP, with YouTube integration.

What you get —

Remote screen control — display media, send notifications, and control playback across multiple displays
YouTube integration — search and play videos on the screen
HTTP and MCP command endpoints, plus MCP-proxy support for clients without SSE (e.g. Claude Desktop)
Langflow setup guidance

Requirements — A YOUTUBE_API_KEY (Google Console, "YouTube Data API v3") and an SS_SERVICE_TOKEN from SmartScreen; a smartscreen.tv device. Python 3.12+.

Cost snapshot — Free and open source under the MIT license; it fronts Google's YouTube Data API (free tier with quotas) and the SmartScreen service, both of which need their own accounts/tokens.

Setup effort — Self-host via uvicorn or Docker; configure a .env with the two tokens.

Our take — Handy for AI-driven digital signage, but only useful if you run SmartScreen displays and set up a Google YouTube API key.

Source: the project README — summarized 2026-07-08.

⌁ readme from github — mirrored (latest on GitHub ↗)

drawing

ultra/mcp-ss is a FastAPI-based MCP server that integrates with smartscreen.tv, a web display service, allowing you to programmatically manipulate the screen (e.g., display media, send notifications, control playback) via simple HTTP/MCP commands.

What is SmartScreen?
Prerequisites
Configuration
Running Locally
Docker
API Endpoints
- Health & Search
- SmartScreen Commands (HTTP)
MCP Tool Integration
Using MCP Proxy for Clients Without SSE Support
Setting up MCP-SS in Langflow
Contributing

mcp logo

What is SmartScreen?

SmartScreen is a web-based screening service.
Content across multiple displays and locations can be controlled remotely. Simply add the MCP tool to your AI app.

SmartScreen setup:

Access SmartScreen through URL: smartscreen.tv
- On Linux, SmartScreen can be installed via snapcraft.
Click the menu button on the top-left corner to reach Settings. Here you can name your device, add tags, and set up OnStart and other scheduled events.
- Don’t forget to click “Save” after making any changes.
Click SmartScreen to return to the Home Page.

ultra/mcp-ss

Docker

docker build -t ultra-mcp-ss .
docker run -d --name ultra-mcp-ss -p 127.0.0.1:8000:8000 ultra-mcp-ss

API Endpoints

Base URL: http://127.0.0.1:8000

Health & Search

HEAD /mcp – health check for MCP
GET /search-youtube?query=... – returns the most relevant YouTube video URL

SmartScreen Commands (HTTP)

Endpoint	Description
POST /drop	Drop media URL onto a screen frame
POST /notify	Send a notification banner
POST /toast	Show a toast popup message
POST /marquee	Display scrolling marquee text
POST /text	Render static text overlay
POST /app	Launch a web app in a frame
POST /touch	Send playback/control commands
POST /status	Query or set system status
POST /dj	Execute DJ tasks: scheduling, kiosk, restart, logo

Refer to OpenAPI docs at http://<host>:8000/docs for request/response schemas.

MCP Tool Integration

FastApiMCP automatically mounts all endpoints as MCP tools under /mcp.
Use your MCP client to invoke tools by name (e.g., drop, notify, toast, etc.).

Using MCP Proxy for Clients Without SSE Support (Claude Desktop)

Install mcp-proxy:

uv pip install --user mcp-proxy #for Python
npm install -g mcp-proxy #for Node.js
pnpm add -g mcp-proxy #for Node.js

On Windows:
Edit claude_desktop_config.json:

{
  "mcpServers": {
    "ultra-mcp-ss": {
      "command": "mcp-proxy",
      "args": ["http://0.0.0.0:8000/mcp"]
    }
  }
}

On MacOS:
Get the path to mcp-proxy:

which mcp-proxy

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "ultra-mcp-ss": {
      "command": "/YOUR/PATH/TO/mcp-proxy",
      "args": ["http://0.0.0.0:8000/mcp"]
    }
  }
}

Setting up MCP-SS in Langflow

To integrate ultra-mcp-ss with Langflow:

drawing

Add MCP Server component from the Tool section in Langflow
Enable Tool Mode in the component settings
Select SSE Mode for real-time communication
Enter the MCP SSE URL: http://0.0.0.0:8000/mcp
Use http://127.0.0.1:8000 if running using Docker.
Run the component to establish connection

Once connected, you can use all SmartScreen commands within your Langflow workflows.

Copy & paste — that's it

{
     "mcpServers": {
       "ultra-mcp-ss": {
         "command": "mcp-proxy",
         "args": ["http://0.0.0.0:8000/mcp"]
       }
     }
   }

Prerequisites

Python 3.12+
Docker (optional, for containerized deployment)
YOUTUBE_API_KEY set up from Google Console for "YouTube Data API v3"
SS_SERVICE_TOKEN environment variable

Configuration

Create a .env file or export environment variables:

YOUTUBE_API_KEY – your Google YouTube Data API v3 key
SS_SERVICE_TOKEN – SmartScreen service token

Example .env:

YOUTUBE_API_KEY=AIzaSy...
SS_SERVICE_TOKEN=xxxxx

or export them:

export YOUTUBE_API_KEY=AIzaSy...
export SS_SERVICE_TOKEN=xxxxx

Running Locally

uvicorn main:app --reload --host 0.0.0.0 --port 8000

Table of Contents