NeoSearch - AI-Powered Search Engine

NeoSearch is a 100% open source, AI-powered search engine that enhances web search with intelligent summaries, multiple perspectives, and real-time streaming results.

Built with ASP.NET Core 9.0 (.NET 9), C# 13, Dapper, SQL Server, and a Vue Petite frontend.

🌟 Features

• 100% Private & Ad-Free: No profiling, no ads, no cookies, no IP logs
• Bias-aware Ranking: AI re-ranks results to prioritize authoritative, independent sources
• Clarity, Instantly: AI rewrites cluttered page titles and descriptions
• Perspectives: Organizes links into logical groups for quick navigation
• AI Overviews: Get quick AI summaries alongside full results
• Smarter Answers & Plugins: Instant results for common queries (weather, conversions, etc.)
• Universal Results: Transforms single-service links into multi-platform results
• Real-time Streaming: Results stream as they're generated via Server-Sent Events

🚀 Quick Start

Prerequisites

• .NET 9 SDK
• SQL Server (or SQL Server Express)
• API keys for services (see Configuration section)

Installation

1. Clone the repository

git clone https://github.com/bartjellema/NeoSearch.git
cd NeoSearch

2. Restore dependencies

dotnet restore

3. Configure the application

Copy appsettings.json.example to appsettings.json and add your API keys:

{
  "ConnectionStrings": {
    "DefaultConnection": "Server=localhost;Database=NeoSearch;Trusted_Connection=true;TrustServerCertificate=True;"
  },
  "APIKeys": {
    "GoogleSearch": "your-google-search-api-key",
    "OpenAI": "your-openai-api-key",
    // Add other API keys as needed
  }
}

4. Run database migrations Sorry, the database schema creation script is coming soon.

5. Build and run

dotnet build
dotnet run --project Server

6. Access the application Open https://localhost:5001 in your browser.

🏗️ Architecture

NeoSearch is built with a modular, microservices-inspired architecture:

Core Components

• Server/ - Main ASP.NET Core application
• APIs/ - External service integrations (YouTube, Spotify, IMDB, etc.)
• LLM/ - LLM provider routing and response generation
• WebSearchService/ - Google search API integration
• Utils/ - Shared utilities and helpers
• ResultHandlers/ - URL pattern-based result enhancers
• Tests/ - Plugin testing with HTTP caching system

Key Patterns

• Parallel Execution: Search tasks run in parallel for optimal performance
• Streaming: Results stream via SSE as they're generated
• Auto-registration: Components auto-register via reflection
• Plugin System: Extensible plugin architecture for specialized features

🔧 Configuration

Required API Keys

Create an account and obtain keys for:

• Google Custom Search API - Core search functionality
• OpenAI API (or alternative LLM providers) - AI features
• Optional: YouTube, Spotify, IMDB, Wikipedia, etc. for enhanced results

🧪 Testing

Run all tests:

dotnet test

Run specific test suites:

dotnet test --filter "FullyQualifiedName~StockPricePluginTests"
dotnet test --filter "FullyQualifiedName~YouTubeTests"

Note: The test suite uses HTTP caching - first run hits real APIs, subsequent runs use cache.

📚 Documentation

• Architecture Overview - Detailed search flow explanation
• API Integration Guide - Adding new API integrations
• Plugin Development - Creating custom plugins
• Result Handlers - Enhancing search results

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

📜 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

🔗 Links

• Website
• Documentation
• Discord Community
• GitHub Repository

📧 Contact

• Founder: Bart Jellema
• Email: b@neosearch.org