RuVector Skill

Self-learning vector database with Graph Neural Networks for autonomous AI memory

Overview

RuVector is a distributed vector database that learns from every query. Unlike static vector databases, RuVector uses GNN (Graph Neural Network) layers to improve search results over time. It's perfect for building self-improving AI memory systems.

When to Use

Use this skill when you need:

• Local vector storage without external API dependencies
• Self-improving memory that gets smarter with usage
• Graph queries with Cypher syntax
• Local LLM integration for RAG without cloud APIs
• Autonomous AI agents that learn from interactions

Key Features

🧠 Self-Learning Index

• GNN layers learn from every query
• Search results improve over time
• No manual index rebuilding needed

🔍 Graph Queries (Cypher)

MATCH (a)-[:SIMILAR]->(b) WHERE a.name = "AI" RETURN b

💾 Local Embeddings

• Built-in ONNX embedding models
• No API calls needed
• Runs entirely offline

⚡ MCP Tools

• 213+ MCP tools for swarm management
• Memory integration
• GitHub automation

Installation

# Quick start
npx ruvector

# Initialize self-learning hooks
npx @ruvector/cli hooks init

# Install optional GNN module
npx ruvector install gnn

Usage Patterns

Basic Vector Storage

const ruvector = require('ruvector');

// Create collection
await db.createCollection('memories', { dimension: 384 });

// Add embeddings
await db.insert('memories', {
  id: 'memory_1',
  vector: embedding,
  metadata: { context: 'user_preference', topic: 'coffee' }
});

// Search (improves over time!)
const results = await db.search('memories', queryEmbedding, { topK: 5 });

Self-Learning Hook

// Enable learning from queries
await db.hooks.enable('self-learning', {
  algorithm: 'q-learning',
  memorySize: 10000
});

Local LLM Integration

// Run LLMs locally
const { RuvLLM } = require('@ruvector/ruvllm');
const llm = new RuvLLM({ model: 'ruvltra-small' });

const response = await llm.chat('Explain vector databases');

Integration with 1ai-skills

RuVector integrates perfectly with:

• runtime-self-improvement - Store learned patterns
• ai-research-agent - Long-term memory
• skill-performance-monitor - Track skill usage

Files in This Skill

• SKILL.md - This file
• references/ - Additional documentation

See Also

• RuVector GitHub
• RuVector NPM
• Documentation