README.md

metadata

title: LoL Multi-Agent Coach
emoji: 🎮
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 6.2.0
app_file: app.py
pinned: false
license: mit
short_description: AI-powered League of Legends coaching system

League of Legends Multi-Agent Coach 🎮

A sophisticated multi-agent AI coaching system for League of Legends that provides:

• 🎯 Match analysis and performance insights
• 🛠️ Build recommendations with real-time meta data
• 🎬 Video guides and educational content
• 📚 LoL knowledge base with FAISS vector search
• 🎲 Pregame strategy (bans, picks, team comp analysis)

Features

• 5 Specialized AI Agents working collaboratively
• 14 Tools for comprehensive coaching
• Real-time data from Riot API, Tavily, and YouTube
• FAISS Vector Database for knowledge retrieval

Setup

This Space requires three API keys to be set as Secrets:

1. OPENAI_API_KEY - Get from https://platform.openai.com/
2. RIOT_API_KEY - Get from https://developer.riotgames.com/
3. TAVILY_API_KEY - Get from https://tavily.com/

Optional: Configure your summoner information:

• SUMMONER_NAME - Your League summoner name
• SUMMONER_TAG - Your tagline (e.g., NA1)
• REGION - Your region (e.g., na1)

Usage

Simply type your questions in the chat interface:

• "Analyze my last 5 games"
• "What's the best build for Yasuo?"
• "Find Yasuo vs Zed matchup videos"
• "What should I ban if playing mid?"
• "Analyze this team comp: Darius, Lee Sin, Ahri, Jinx, Thresh"

Technology Stack

• LangChain/LangGraph - Multi-agent framework
• OpenAI GPT-4 - Language model
• Gradio 6.2.0 - Web interface
• FAISS - Vector database (optional, see FAISS_SETUP.md)
• Riot API - Live game data
• Tavily - Real-time web search

Note: FAISS Knowledge Base

The FAISS knowledge base is optional and provides enhanced LoL knowledge retrieval. If not available, the system will work perfectly using Tavily web search instead. See FAISS_SETUP.md for upload instructions if you want to enable it.