Livetennisapi MCP Integration

livetennisapi-mcp Integration

Skill by ara.so — MCP Skills collection.

The livetennisapi-mcp server provides AI agents with access to real-time tennis data through the Model Context Protocol (MCP). It exposes 12 tools for querying live scores, upcoming matches, player profiles, odds, and ML-powered win probabilities across ATP, WTA, Challenger, and ITF tournaments.

Installation

Get an API Key

1. Visit livetennisapi.com/subscribe/free for a free key (no credit card)
2. Store your key in an environment variable:

export LIVETENNISAPI_KEY=twjp_your_actual_key_here

Claude Code

claude mcp add livetennis -e LIVETENNISAPI_KEY=twjp_your_key -- npx -y livetennisapi-mcp

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "livetennis": {
      "command": "npx",
      "args": ["-y", "livetennisapi-mcp"],
      "env": { "LIVETENNISAPI_KEY": "twjp_your_key_here" }
    }
  }
}

Config file locations:

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

Cursor / Zed

Same configuration pattern as Claude Desktop. Add the server block to your MCP settings file.

Codex

codex mcp add livetennisapi \
  --url https://mcp.livetennisapi.com/mcp \
  --bearer-token-env-var LIVETENNISAPI_KEY

Or manually in ~/.codex/config.toml:

[mcp_servers.livetennisapi]
url = "https://mcp.livetennisapi.com/mcp"
bearer_token_env_var = "LIVETENNISAPI_KEY"

Available Tools

FREE Tier

BASIC Tier

• get_recent_results — Completed matches and winners

PRO Tier

• get_match_events — Breaks, games, sets, momentum shifts
• get_match_odds — Match-winner betting prices (bid/ask/mid)

ULTRA Tier

• get_match_analysis — ML model win probability and key factors

Utility

• check_api_status — Verify API connectivity and key tier

Usage Patterns

Checking Live Matches

User prompt:

What tennis matches are live right now?

The agent will call get_live_matches and receive structured data:

{
  "matches": [
    {
      "match_id": 18953,
      "tournament": "Australian Open",
      "round": "QF",
      "player1": "Carlos Alcaraz",
      "player2": "Jannik Sinner",
      "score": "6-3, 2-4",
      "serving": 2,
      "status": "live"
    }
  ]
}

Player Lookup

User prompt:

Show me Sinner's ranking and recent results

The agent will:

1. Call search_players with query "Sinner"
2. Call get_player with the player_id
3. Call get_recent_results (if key has BASIC+)

{
  "player_id": 12345,
  "name": "Jannik Sinner",
  "country": "ITA",
  "ranking": 1,
  "ranking_points": 11830,
  "handedness": "right",
  "age": 23
}

Match Details with Odds

User prompt:

What are the odds on the Alcaraz vs Sinner match?

For PRO+ keys, the agent calls:

1. get_live_matches or get_match to find match_id
2. get_match_odds with that match_id

{
  "match_id": 18953,
  "player1_odds": {
    "bid": 1.85,
    "ask": 1.90,
    "mid": 1.875
  },
  "player2_odds": {
    "bid": 1.95,
    "ask": 2.00,
    "mid": 1.975
  }
}

Win Probability Analysis

User prompt:

What does the model give Alcaraz in this match?

For ULTRA keys, the agent calls get_match_analysis:

{
  "match_id": 18953,
  "player1_win_probability": 0.48,
  "player2_win_probability": 0.52,
  "key_factors": [
    "Sinner's serve hold rate on hard courts: 89%",
    "Alcaraz's recent injury recovery affecting movement"
  ],
  "model_thesis": "Close match. Sinner slight favorite due to surface advantage."
}

Tier-Aware Responses

The server returns human-readable explanations instead of raw 403 errors when a tool requires a higher tier:

This data requires the PRO plan, and the configured API key is on FREE tier.
Upgrade at https://livetennisapi.com/#pricing

This prevents the agent from hallucinating reasons or retrying endlessly.

Check Your Key's Tier

What tier is my API key on?

The agent calls check_api_status, which probes upward to determine:

{
  "status": "reachable",
  "tier": "PRO",
  "message": "API is reachable. Your key is on the PRO plan."
}

HTTP Endpoint (Alternative)

For clients that cannot run stdio MCP servers, use the hosted HTTP endpoint:

https://mcp.livetennisapi.com/mcp

With Claude Messages API (Python)

import os
import anthropic

client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

response = client.beta.messages.create(
    model="claude-opus-4-8",
    max_tokens=4096,
    betas=["mcp-client-2025-11-20"],
    mcp_servers=[{
        "type": "url",
        "name": "livetennisapi",
        "url": "https://mcp.livetennisapi.com/mcp",
        "authorization_token": os.environ["LIVETENNISAPI_KEY"],
    }],
    tools=[{"type": "mcp_toolset", "mcp_server_name": "livetennisapi"}],
    messages=[{
        "role": "user",
        "content": "What tennis matches are live right now?"
    }],
)

print(response.content)

With Claude Web Connector

Add a custom connector in Claude with:

https://mcp.livetennisapi.com/mcp?token=twjp_your_key

Security note: Prefer Authorization: Bearer header over ?token= when your client supports it. The query param exists for clients that cannot set headers.

Development & Self-Hosting

Local Development

git clone https://github.com/livetennisapi/livetennisapi-mcp.git
cd livetennisapi-mcp
npm install
npm run build

# Run stdio server
LIVETENNISAPI_KEY=twjp_your_key node dist/index.js

# Run HTTP server (port 8081)
LIVETENNISAPI_KEY=twjp_your_key node dist/http.js

Testing

npm test               # Protocol + transport isolation + rate limiting
npm run test:mutation  # Proves tests fail when code breaks

The mutation tests are important: they reintroduce bugs the tests claim to catch and verify the suite actually detects them.

Self-Host HTTP Endpoint

See deploy/install-http.sh and deploy/TUNNEL.md in the repository for instructions on running your own HTTP endpoint.

The hosted endpoint is multi-tenant and holds no key of its own — every request builds an isolated server bound to the key provided in that request.

Troubleshooting

"Server not found" or "Connection refused"

Stdio version:

• Ensure npx can reach npm registry: npx -y livetennisapi-mcp --version
• Check Node.js version: node --version (needs 18+)
• Restart your AI client after adding the server config

HTTP version:

• Verify endpoint is reachable: curl https://mcp.livetennisapi.com/mcp
• Check firewall/proxy settings

"Invalid API key" or 401 errors

# Verify your key is set correctly
echo $LIVETENNISAPI_KEY

# Test key directly against API
curl -H "Authorization: Bearer $LIVETENNISAPI_KEY" \
  https://api.livetennisapi.com/v1/live_matches

If the curl works but MCP doesn't, the env var isn't reaching the server process. In Claude Desktop, ensure env is in the server config, not at the root level.

"upgrade_required" responses

Your key works, but the endpoint requires a higher tier. Use check_api_status to see your current tier, then upgrade at pricing.

Rate limiting

• Anonymous requests to HTTP endpoint: 60/minute
• Keyed requests to HTTP endpoint: 300/minute
• Per-key quotas enforced upstream by tier

If you hit the limit, the server returns a 429 with Retry-After header.

No live matches returned

Tennis is not 24/7. Grand Slams and peak season provide the most coverage. If get_live_matches returns an empty array, try get_upcoming_matches or get_fixtures to see the schedule.

Code Examples

TypeScript: Using the Underlying Client

The MCP server is built on the official livetennisapi client. You can use it directly:

import { LiveTennisAPI } from 'livetennisapi';

const client = new LiveTennisAPI(process.env.LIVETENNISAPI_KEY!);

// Get live matches
const liveMatches = await client.getLiveMatches();
console.log(liveMatches);

// Search for a player
const players = await client.searchPlayers('Djokovic');
const playerId = players[0]?.player_id;

// Get player details
if (playerId) {
  const player = await client.getPlayer(playerId);
  console.log(`${player.name} is ranked #${player.ranking}`);
}

// Get match odds (PRO tier)
const odds = await client.getMatchOdds(18953);
console.log(`Player 1 mid: ${odds.player1_odds.mid}`);

JavaScript: Programmatic MCP Tool Calls

import { Client } from '@modelcontextprotocol/sdk/client/index.js';
import { StdioClientTransport } from '@modelcontextprotocol/sdk/client/stdio.js';

const transport = new StdioClientTransport({
  command: 'npx',
  args: ['-y', 'livetennisapi-mcp'],
  env: { LIVETENNISAPI_KEY: process.env.LIVETENNISAPI_KEY }
});

const client = new Client({
  name: 'tennis-client',
  version: '1.0.0'
}, { capabilities: {} });

await client.connect(transport);

// List available tools
const { tools } = await client.listTools();
console.log(tools.map(t => t.name));

// Call a tool
const result = await client.callTool({
  name: 'get_live_matches',
  arguments: {}
});

console.log(result.content);

Related Projects

• Python client: pip install livetennisapi — livetennisapi-python
• JavaScript client: npm install livetennisapi — livetennisapi-js
• Vercel AI SDK tools: npm install livetennisapi-ai — livetennisapi-ai
• OpenAPI spec: livetennisapi/openapi

Resources

• API Documentation: docs.livetennisapi.com
• Pricing & Plans: livetennisapi.com/#pricing
• Terms of Service: livetennisapi.com/terms
• Support: Open an issue at github.com/livetennisapi/livetennisapi-mcp