Sports MCP Server for AI Agents

Endpoint

GET /mcp returns the discovery document as JSON — that is intentional. This HTML page is the SEO/human landing for “sports MCP server.” Tool calls are POST only; some clients briefly probe with Accept: text/event-stream and receive 405 Allow: POST before continuing over POST JSON.

Auth

Pass a Lumify API key as a Bearer token on every MCP request:

Authorization: Bearer YOUR_API_KEY

Fastest start: instant trial key at /docs/ai (100 credits, 14 days, no signup). Persistent Free Tier: /register (1,000 credits).

Tools

18 tools map 1:1 to REST. Credit cost mirrors the equivalent HTTP call. Live schemas always come from tools/list.

Clients & config

Cursor (remote — recommended)

{
  "mcpServers": {
    "lumify": {
      "url": "https://lumify.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

One-click install (replace the placeholder key before approving): Add Lumify MCP to Cursor.

Claude Desktop (stdio bridge)

{
  "mcpServers": {
    "lumify": {
      "command": "npx",
      "args": ["-y", "@lumifyai/mcp"],
      "env": { "LUMIFY_API_KEY": "YOUR_API_KEY" }
    }
  }
}

VS Code / Copilot

{
  "servers": {
    "lumify": {
      "type": "http",
      "url": "https://lumify.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

More recipes: /docs/guides#mcp.

Example invocation

After connect, agents typically: estimate_cost → list_events → get_odds / get_intelligence. Equivalent REST for local testing:

curl -s -X POST https://lumify.ai/v1/estimate \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"calls":[{"tool":"get_odds","arguments":{"event_id":4821,"bookmaker":"all"}}]}'

curl -s "https://lumify.ai/v1/events/4821/odds?bookmaker=all" \
  -H "Authorization: Bearer YOUR_API_KEY"

from lumify import Lumify

client = Lumify(api_key="YOUR_API_KEY")
client.estimate.cost([{
    "tool": "get_odds",
    "arguments": {"event_id": 4821, "bookmaker": "all"},
}])
odds = client.events.odds(4821, bookmaker="all")
print(odds["available"], len(odds.get("bookmakers") or []))

import { Lumify } from "@lumifyai/sdk";

const client = new Lumify({ apiKey: "YOUR_API_KEY" });
await client.estimate.cost([{
  tool: "get_odds",
  arguments: { event_id: 4821, bookmaker: "all" },
}]);
const odds = await client.events.odds(4821, { bookmaker: "all" });
console.log(odds.available, odds.bookmakers?.length);

Over MCP, the same get_odds tool returns identical JSON plus _meta.credits_used.

Sports & books

Same coverage as the REST Sports Data / Odds APIs: major US + global sports for schedules and scores; nine sportsbooks for moneyline / spread / totals; intelligence live for MLB, NFL, NCAAF, tennis, FIFA WC soccer, and MLS (others may be data+odds only).

Deep dives: Sports Data API · Sports Odds API · API for AI agents.

Billing

• Handshake free: initialize, tools/list, ping
• Metered: tools/call only — same credits as REST
• Budget first: estimate_cost (always free)