Al Jazeera 360 MCP Server

Connect any AI assistant to Al Jazeera 360's streaming catalog — search, browse, and retrieve Arabic video content with direct watch links.

An MCP (Model Context Protocol) server that gives AI tools like Claude, ChatGPT, Gemini, and Cursor real-time access to Al Jazeera 360's full content library: documentaries, investigative programs, talk shows, podcasts, and original productions.

⚠️ Unofficial community project. Not affiliated with, endorsed by, or sponsored by Al Jazeera Media Network. It uses the platform's publicly accessible API and links back to aljazeera360.com for all content.

⚡ Quick Connect — No Install Needed

A hosted instance is live on Cloudflare. Add this to your MCP client config and you're done:

{
  "mcpServers": {
    "aljazeera360": {
      "type": "http",
      "url": "https://aljazeera360-mcp.ahmed-26d.workers.dev/mcp"
    }
  }
}

First request after idle takes a few seconds (container cold start) — subsequent requests are fast.

Then ask your assistant: "What's trending on Al Jazeera 360?" — or in Arabic: "ايه أحدث حلقات الدحيح؟"

What It Does

Ask your AI assistant questions like:

• "What's trending on Al Jazeera 360?"
• "Find documentaries about Palestine"
• "Show me the latest episodes of Al Daheeh"
• "Search for videos about Gaza"

The AI connects to this server, fetches real data from Al Jazeera 360, and returns actual video titles, descriptions, durations, and direct watch links to aljazeera360.com.

Tools Available

The server ships with two tool profiles:

A small default toolset keeps AI tool selection fast and accurate. Enable the full profile only if you need the SEO/analytics tools.

Core Discovery Tools (always on)

SEO & Metadata Tools (requires AJ360_ENABLE_SEO_TOOLS=1)

Advanced AI & Analytics Tools (requires AJ360_ENABLE_SEO_TOOLS=1)

Sections

Quick Start

Install

git clone https://github.com/ahmedaminsa/aljazeera360-mcp-server.git
cd aljazeera360-mcp-server
pip install -r requirements.txt

Configure Authentication

The server supports multiple authentication methods (in priority order):

1. Refresh Token (Recommended) — Auto-refreshes every 10 minutes, valid for ~1 year:

   server.ts

   Copy

   export AJ360_REFRESH_TOKEN="eyJ..."
   

2. Auth Token — Direct token, expires in ~10 minutes:

   server.ts

   Copy

   export AJ360_AUTH_TOKEN="eyJ..."
   

3. Guest Mode — No configuration needed. The server auto-creates a guest session.

How to get your refresh token:

1. Open https://www.aljazeera360.com in your browser
2. Log in with your account
3. Open DevTools → Console → Run: localStorage.getItem('dice:refreshToken')
4. Copy the token value

Test

python test_server.py

Core tools are tested against the live API.

Run

python server.py

Use with Claude Desktop

Add to your config file:

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

{
  "mcpServers": {
    "aljazeera360": {
      "command": "python",
      "args": ["/full/path/to/aljazeera360-mcp-server/server.py"],
      "env": {
        "AJ360_REFRESH_TOKEN": "your-refresh-token-here"
      }
    }
  }
}

Restart Claude Desktop. The Al Jazeera 360 tools will appear automatically.

Use with Any MCP Client

This server speaks the standard MCP protocol over stdio (local) and Streamable HTTP (hosted). It works with any MCP-compatible client:

• Claude Desktop
• Cursor
• Continue
• Windsurf
• MCP Inspector (for testing)
• Any custom MCP client

Configuration

The server needs only AJ360_API_KEY to run — authentication falls back to an automatic guest session. For full content access, also provide a refresh token.

Deploy

The production instance runs on Cloudflare Containers — full walkthrough in deploy/cloudflare/README.md (wraps the Dockerfile in a Cloudflare Container behind a tiny Worker; requires the Workers Paid plan and local Docker for deploys). Any container platform works — alternatives below.

Docker

docker build -t aljazeera360-mcp .
docker run -p 8080:8080 -e MCP_TRANSPORT=streamable-http -e AJ360_REFRESH_TOKEN=your-token aljazeera360-mcp

Google Cloud Run

gcloud builds submit --tag gcr.io/YOUR_PROJECT/aljazeera360-mcp
gcloud run deploy aljazeera360-mcp \
  --image gcr.io/YOUR_PROJECT/aljazeera360-mcp \
  --platform managed \
  --allow-unauthenticated \
  --set-env-vars="MCP_TRANSPORT=streamable-http,AJ360_REFRESH_TOKEN=your-token"

Render / Railway / Fly.io

Connect this repo — auto-deploys from the included Dockerfile. Set environment variables in the dashboard (including AJ360_ALLOWED_HOST=<your-domain> so the DNS-rebinding protection accepts your host).

How It Works

Your AI Assistant
       │
       │  MCP Protocol (stdio / Streamable HTTP)
       ▼
┌──────────────────────────┐
│  This MCP Server         │
│  ┌────────────────────┐  │
│  │ Token Manager      │  │
│  │ (auto-refresh)     │  │
│  ├────────────────────┤  │
│  │ 24 Tools + Prompts │  │
│  │ + Retry + Cache    │  │
│  └────────────────────┘  │
└──────────┬───────────────┘
           │
           │  HTTPS (authenticated)
           ▼
┌──────────────────────────┐
│  Al Jazeera 360 API      │
│  (Vesper/Dice Platform)  │
└──────────┬───────────────┘
           │
           ▼
┌──────────────────────────┐
│  aljazeera360.com        │
│  (Direct watch links)    │
└──────────────────────────┘

Built-in Prompts

The server includes pre-built prompts for common AI scenarios:

Example Output

Calling search_videos("غزة") returns:

{
  "query": "غزة",
  "total_results": 20,
  "results": [
    {
      "title": "المسعفون أهداف إسرائيل المتنقلة!",
      "type": "VOD",
      "duration": "24:53",
      "id": "965741",
      "watch_url": "https://www.aljazeera360.com/video/965741"
    },
    {
      "title": "قيادة السنوار لجهاز حماس الأمني",
      "type": "VOD",
      "duration": "54:57",
      "id": "953659",
      "watch_url": "https://www.aljazeera360.com/video/953659"
    }
  ]
}

Analytics Dashboard

The server includes a built-in analytics dashboard that tracks every request made by AI tools.

What It Tracks

• Which AI tool made the request (Claude, ChatGPT, Cursor, etc.)
• Which MCP tool was called (search_videos, get_trending_content, etc.)
• When each request happened (timestamp)
• How long each request took (response time in ms)
• Whether it succeeded or failed
• Top search terms (what users are asking about)
• Daily/hourly activity patterns

Access the Dashboard

The dashboard starts automatically on port 9090 when you run the server:

http://localhost:9090

For cloud deployments, expose port 9090 alongside the MCP port (8080).

API Endpoints

Configuration

Example Stats Output

{
  "total_requests": 1247,
  "success_rate": "99.2%",
  "tools_usage": [
    {"tool": "search_videos", "calls": 523, "avg_response_ms": 340},
    {"tool": "get_trending_content", "calls": 312, "avg_response_ms": 180}
  ],
  "clients": [
    {"client": "claude-desktop", "requests": 890},
    {"client": "cursor", "requests": 357}
  ],
  "top_searches": [
    {"term": "\u063a\u0632\u0629", "count": 89},
    {"term": "\u0641\u0644\u0633\u0637\u064a\u0646", "count": 67}
  ]
}

Tech Stack

License

MIT

SEO Ops & Vesper Knowledge Base (AI-assisted)

This repo doubles as an AI-assisted SEO operations workspace for the platform. Open it in Claude Code and you get:

export AJ360_API_KEY=…
python scripts/seo_snapshot.py   # → reports/seo/<date>.md (local only)

Contributing

PRs welcome. See CONTRIBUTING.md for guidelines.

Run tests before submitting:

python test_server.py