valorant-mcp

MCP server for Valorant esports — match schedules, live results, reminders, and AI-powered analysis.

Any MCP-compatible AI agent (Claude, Cursor, Windsurf, etc.) can install this and immediately ask questions like "What matches are live right now?", "Remind me before Sentinels' next match", or "Who was MVP in yesterday's Champions final?"

Data is sourced from vlr.gg. Analysis is powered by Claude (optional).

Quickstart

# Run directly (no install)
npx valorant-mcp

# Or install globally
npm install -g valorant-mcp

Add to Claude Desktop / Claude Code

// ~/.claude/claude_desktop_config.json
{
  "mcpServers": {
    "valorant": {
      "command": "npx",
      "args": ["valorant-mcp"],
      "env": {
        "ANTHROPIC_API_KEY": "your-key-here"
      }
    }
  }
}

ANTHROPIC_API_KEY is only required for get_match_analysis and get_match_preview. All other tools work without it.

Related MCP server: League of Legends Stats MCP Server

Tools

Example Conversations

"What Valorant matches are happening today?"

"Remind me 10 minutes before Sentinels' next match"

"Who won the Champions 2025 final and what was the score?"

"Give me a full analysis of the last Team Liquid match"

"What matches are live right now?"

Environment Variables

Pi Daemon (background reminders)

The MCP server handles on-demand queries from AI agents. For automatic notifications without needing an agent open, run the Python daemon on a Raspberry Pi or any always-on device:

pip install requests

export TELEGRAM_BOT_TOKEN="your_token"
export TELEGRAM_CHAT_ID="your_chat_id"

python3 daemon/valorant_daemon.py

The daemon reads reminders saved by the MCP server (subscribe_reminder tool) and fires notifications when a match is about to start.

Auto-start on boot

# Edit the service file with your credentials
nano daemon/valorant-daemon.service

sudo cp daemon/valorant-daemon.service /etc/systemd/system/
sudo systemctl enable --now valorant-daemon
sudo systemctl status valorant-daemon

Build from Source

git clone https://github.com/YOUR_USERNAME/valorant-mcp
cd valorant-mcp
npm install
npm run build
npm start

# Test with the MCP Inspector UI
npm run inspector

Architecture

src/
├── index.ts              # MCP server entry point (9 tools)
├── scraper/
│   ├── vlr.ts            # vlr.gg scraper
│   └── types.ts          # Shared TypeScript interfaces
├── tools/
│   ├── matches.ts        # Match listing and search tools
│   ├── analysis.ts       # AI analysis and preview tools
│   └── reminders.ts      # Reminder management tools
├── analysis/
│   ├── claude.ts         # Anthropic API client
│   └── prompts.ts        # Analysis and preview prompt templates
└── db/
    └── index.ts          # SQLite (reminders + response cache)

daemon/
├── valorant_daemon.py    # Python background worker for Pi
└── valorant-daemon.service  # systemd service file

How It Works

1. Scraping — Match data is scraped from vlr.gg on demand. Results are cached for 1 hour.

2. Reminders — Stored in a local SQLite DB (~/.valorant-mcp/data.db). The Pi daemon polls every 60 seconds and fires notifications for due reminders.

3. Analysis — Match stats are fed to Claude which returns a structured natural-language breakdown. Results are cached permanently (no re-billing for the same match).

Where to Find MCP Servers

Looking for more MCP servers? Check these directories:

• Smithery — largest MCP marketplace

• mcp.so — community directory

• Glama — curated list

• awesome-mcp-servers — GitHub list

License

MIT