MCP server overview

The Model Context Protocol (MCP) is an open standard for connecting AI agents to data sources. Scorelytics ships an MCP server that exposes the same data as the REST API, but in a way LLMs can call directly — no glue code, no custom integration.

Endpoint

https://api.scorelytics.pro/v1/mcp

The server speaks JSON-RPC 2.0 over HTTP (Streamable HTTP transport, spec 2025-03-26).

It supports POST for tool calls and GET for the SSE stream required by the spec (no server-initiated notifications today; the GET responds with a single confirmation event).

What you can do with it

The server exposes eight tools that map 1-to-1 to the major REST endpoints:

Tool	Returns
list_leagues	All tracked leagues.
get_matches	Filtered match list (date / league / status).
get_match	Full match record with localized period label.
get_match_events	Chronological events.
get_match_stats	In-game statistics.
get_match_lineups	Starting XI + bench.
get_standings	Current league table.
get_head_to_head	History between the two teams of a match.

Full tool reference →

Quick test from the CLI

curl https://api.scorelytics.pro/v1/mcp \
  -H "Authorization: Bearer sk_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/list"
  }'

The response lists every tool with its input schema.

Auth

The MCP endpoint accepts the same three formats as the REST API:

• Authorization: Bearer sk_live_...
• X-API-Key: sk_live_...
• ?api_key=sk_live_...

For Claude.ai, the endpoint also implements the OAuth 2.0 Authorization Server discovery flow (RFC 9728 + RFC 7591) so users can connect with a button click instead of pasting a key. See the OAuth integration page.

Connecting Claude.ai

1. Open Claude → Settings → Integrations → MCP.
2. Add https://api.scorelytics.pro/v1/mcp.
3. Sign in with your Scorelytics account when prompted.

Claude will discover the eight tools automatically. From there, ask things like "What's happening in the Premier League right now?" and Claude will call get_matches for you.

Step-by-step Claude integration →

Why MCP over REST?

For an AI agent, MCP is materially better than wrapping REST yourself:

• Discovery is free — the agent learns your tools by calling tools/list. No prompt engineering required.
• Schemas are typed — each tool has a JSON Schema for inputs, so the model produces valid arguments.
• Auth is standardized — Bearer tokens or OAuth, same flow as any other MCP integration the user has.

If you're not building an agent and just need data, the REST API is a better fit.