A powerful, extensible AI assistant platform with multi-model support and advanced memory capabilities.

• 🤖 Multiple LLM Support (Gemini, Ollama)
• 💭 Long-term Memory with Vector Storage
• 🔧 Extensible Tool System
• 💬 Telegram Bot Integration
• 🔄 Automatic Failover Between Models
• 📊 Vector Similarity Search
• 🚀 Rate Limiting & Performance Optimization

• Docker & Docker Compose
• .NET 8.0
• PostgreSQL with pgvector
• Redis
• Ollama running
• Hugging Faces Api Key
• Telegram Bot Api Key
• Gemini Api Key

1. Start Infrastructure: docker-compose up -d

2. Configure Environment:

• Set required API keys and configurations on appsettings.*.json

3. Pull models on ollama

   $ ollama pull qwen2.5:1.5b
   $ ollama pull all-minilm:l6-v2

4. Start the Application: dotnet run --project Api/RootBootstrapper

• API Layer: ASP.NET Core with MediatR
• LLM Integration: Multi-provider system with fallback
• Storage:
  • PostgreSQL + pgvector for embeddings
  • Redis for caching
• Memory System: RAG-based with vector similarity search
• Tools: Modular system for web search and file operations

Key settings in appsettings.json:

{
  "ConnectionStrings": {
    "DefaultConnection": "Host=localhost;Port=5432;Username=postgres;Password=postgres;Database=horusdb;",
    "Redis": "localhost:6380,password=yourpassword,ssl=False"
  },
  "Telegram": {
    "Token": "your-telegram-token"
  },
  "RedisConfig": {
    "ApplicationPrefixKey": "Horus",
    "KeyDelimiter": ":",
    "AuditListKey": "Audit"
  },
  "Gemini": {
    "ApiKey": "your-gemini-key",
    "ModelName": "gemini-1.5-flash",
    "TokensPerSecond": 0.25,
    "Burst": 5
  },
  "Ollama": {
    "Url": "http://localhost:11434",
    "ModelName": "qwen2.5:1.5b"
  },
  "Supabase": {
    "HuggingFaceApiKey": "your-huggingface-key",
    "HuggingFaceModelName": "sentence-transformers/all-MiniLM-L6-v2"
  }
}

• Use Visual Studio or VS Code
• Debug profile available for local development
• Docker support for production deployment

1. Fork the repository
2. Create feature branch
3. Submit pull request

MIT License