⚽ FIFA World Cup 2026 — MCP Server & Dashboard

An MCP server built with FastMCP to query FIFA World Cup 2026 data directly from any AI compatible with the MCP protocol (Claude, Cursor, etc.), plus a Streamlit dashboard for standalone use.

Data is sourced from openfootball/worldcup.json — open source, no API key required.

🚀 Available Tools

Related MCP server: mcp-claudinho

📦 Installation

git clone https://github.com/Eduardo-Lucas/worldcup-2026-mcp
cd worldcup-mcp

pip install fastmcp httpx streamlit

🖥️ Streamlit Dashboard

A visual dashboard with English/Portuguese language switching (🇺🇸 / 🇧🇷):

streamlit run src/app.py

Open http://localhost:8501 in your browser.

Pages:

• 🏆 Recent Matches — scores and goal scorers

• 📅 Upcoming Matches — schedule with venues

• 📊 Standings — all 12 groups with W/D/L/GF/GA/GD

• 🔍 Team Search — full team history and upcoming fixtures

• 📈 Statistics — top scorers, biggest win, goals per game

⚙️ MCP Server — Claude Code

claude mcp add worldcup-2026 python -- -m src.server

⚙️ MCP Server — Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "worldcup-2026": {
      "command": "python",
      "args": ["-m", "src.server"],
      "cwd": "/absolute/path/to/worldcup-mcp"
    }
  }
}

Restart Claude Desktop. The tools will be available automatically.

💬 Usage Examples

Once connected, ask Claude:

• "What were yesterday's World Cup results?"

• "How is Brazil doing in the standings?"

• "Show me Group C table"

• "Who are the top scorers so far?"

• "What are Brazil's upcoming matches?"

🧪 Tests

pip install pytest pytest-asyncio
pytest tests/ -v

🏗️ Project Structure

worldcup-mcp/
├── src/
│   ├── server.py          # MCP server (FastMCP tools)
│   └── app.py             # Streamlit dashboard
├── tests/
│   └── test_server.py     # Tool unit tests
├── pyproject.toml
├── claude_desktop_config.json
└── README.md

📝 How It Works

• @mcp.tool() — exposes Python functions as AI tools

• Docstrings — become the tool description the AI reads

• Type hints — define the parameter schema

• mcp.run() — starts the server over stdio

• Standings — calculated dynamically from match results (no separate endpoint needed)

• Placeholder resolver — knockout stage codes like W73, 1G, 3A/B/C/D/F are automatically resolved to real team names as results come in

Built with FastMCP · Data: openfootball/worldcup.json