LORG

The Search Engine That Thinks Before It Searches

Website • Documentation • Skelf Research

---

LORG is an AI-powered search engine that combines Large Language Models with knowledge graph generation to deliver precise, context-aware answers. Unlike traditional search, LORG understands your query, builds a knowledge graph, and intelligently scores web results for relevance.

Stop searching. Start finding.

Why LORG?

• AI-First Search - Generates answers using GPT models before searching the web
• Knowledge Graphs - Automatically extracts entities and relationships from queries
• Smart Ranking - Uses AI to score results by actual relevance, not just keywords
• Multi-Engine - Aggregates results from Google, Bing, Yahoo & DuckDuckGo
• Flexible - Use as CLI, REST API, Node.js library, or full web app

Quick Start

# Install globally
npm install -g lorg

# Set your OpenAI API key
export OPENAI_API_KEY=sk-your-key-here

# Search!
lorg "What causes the northern lights?"

Output:

Searching for: "What causes the northern lights?"

=== LLM Answer ===
The northern lights (aurora borealis) are caused by charged particles
from the sun colliding with gases in Earth's atmosphere...

=== Knowledge Graph ===
{ entities: ["aurora", "solar wind", "magnetosphere", ...] }

=== Search Results ===
1. Aurora Borealis Explained (Score: 0.94)
   Source: https://science.nasa.gov/aurora

Total tokens used: 1,247

Installation

# Global (for CLI)
npm install -g lorg

# Local (for library use)
npm install lorg

Requirements: Node.js 18+ and an OpenAI API key

Usage

Command Line

# Basic search
lorg "How do black holes form?"

# Use GPT-4 for higher quality
lorg -m gpt-4o "Explain quantum entanglement"

# Limit to 3 results
lorg -r 3 "Best practices for React hooks"

REST API Server

# Start the server
lorg server -p 3000

# Query via curl
curl -X POST http://localhost:3000/search \
  -H "Content-Type: application/json" \
  -d '{"query": "How does CRISPR work?"}'

Node.js Library

import LorgSearch from 'lorg';

const lorg = new LorgSearch(process.env.OPENAI_API_KEY);

const results = await lorg.search('What is quantum computing?', {
  model: 'gpt-4o-mini',
  maxResults: 5
});

console.log(results.answer);        // AI-generated answer
console.log(results.knowledgeGraph); // Extracted entities & relationships
console.log(results.keywords);       // Search keywords
console.log(results.results);        // Ranked web results
console.log(results.tokenCount);     // Tokens consumed

How It Works

Query → LLM Answer → Knowledge Graph → Keywords → Web Search → AI Ranking → Results

1. Answer Generation - GPT generates an initial answer and knowledge graph
2. Keyword Extraction - AI extracts optimal search terms from the graph
3. Multi-Engine Search - Queries Google, Bing, Yahoo & DuckDuckGo via SearxNG
4. Content Analysis - Fetches pages and extracts relevant content
5. Relevance Scoring - AI scores each result (0-1) for query relevance
6. Result Ranking - Returns top results sorted by relevance score

Configuration

Environment Variable	Default	Description
OPENAI_API_KEY	-	Your OpenAI API key (required)
OPENAI_BASE_URL	OpenAI	Custom OpenAI-compatible endpoint
LORG_DEFAULT_MODEL	gpt-4o-mini	Default model for all operations
LORG_SEARCH_ENGINES	google,bing,yahoo,duckduckgo	Search engines to use
LORG_SEARCH_LIMIT	20	Max results from search API

API Reference

new LorgSearch(apiKey?)

Create a new LORG instance. Falls back to OPENAI_API_KEY env var.

search(query, options?)

Option	Type	Default	Description
model	string	gpt-4o-mini	OpenAI model to use
maxResults	number	5	Maximum results to return

Returns:

{
  answer: string;           // AI-generated answer
  knowledgeGraph: object;   // Entities and relationships
  keywords: string[];       // Extracted search keywords
  results: Array<{
    title: string;
    source: string;
    score: number;          // 0-1 relevance score
    content: string;
  }>;
  tokenCount: number;       // Total tokens used
}

Documentation

Full documentation available at docs.skelfresearch.com/slorg

• Getting Started
• CLI Reference
• API Documentation
• Architecture

Use Cases

• Research Assistants - Get accurate answers with sources
• Content Discovery - Find relevant articles ranked by AI
• RAG Pipelines - High-quality retrieval for LLM applications
• Knowledge Base Search - Semantic search over web content
• Fact Checking - Verify claims with scored sources

Contributing

Contributions are welcome! Please see our GitHub repository.

git clone https://github.com/skelf-research/slorg.git
cd slorg
npm install
npm run dev

License

MIT License - see LICENSE for details.

---

Built with AI, for AI-powered applications
GitHub · Documentation · npm

---

Part of Skelf Research

slorg is built by Skelf Research — an independent UK AI research lab publishing production-grade open-source projects.

🌐 Website · 📚 Documentation · 🔬 All projects · 🤗 Hugging Face

Related projects: polymathy (answer-engine service) · memista (vector search for Rust) · embedcache (local embedding cache)

_{Released under MIT / Apache-2.0. © Skelf Research Limited.}