AgentDB vs RuVector - Usage Guide banner
marcuspat marcuspat

AgentDB vs RuVector - Usage Guide

Development community intermediate

Description

> **DEPRECATION NOTICE:** This document was written for Claude Flow V3. In Turbo Flow V4, Ruflo v3.5 bundles both AgentDB and RuVector — there is no separate installation. All `claude-flow` CLI comman

Installation

Terminal
claude install-skill https://github.com/marcuspat/turbo-flow

README

AgentDB vs RuVector - Usage Guide

**DEPRECATION NOTICE:** This document was written for Claude Flow V3. In Turbo Flow V4, Ruflo v3.5 bundles both AgentDB and RuVector — there is no separate installation. All `claude-flow` CLI commands below should be replaced with `npx ruflo@latest` equivalents. See the V4 Quick Reference Guide for current commands. This file is preserved for historical context.

**When to use AgentDB versus RuVector in Claude Flow V3**


🎯 TL;DR - Quick Decision Matrix

Your Need Use This Why
Vector memory storage AgentDB Built into Claude Flow, persistent HNSW indexing
Code intelligence (routing, AST, diff) RuVector ML-based agent routing, code analysis
Distributed swarms (100+ agents) RuVector Postgres Centralized coordination across hosts
Local development (1-15 agents) AgentDB Simpler, already installed
Advanced neural (LoRA, EWC++, Flash Attention) RuVector Native Rust performance
Simple semantic search AgentDB Good enough for most cases

📊 Overview

What They Are

**AgentDB:**

    undefined

**RuVector:**

    undefined

🔍 AgentDB - Built-in Vector Memory Storage

What It Is

Vector database component built into `@claude-flow/memory` that provides:

    undefined

Core Capabilities

import { AgentDBAdapter, HNSWIndex } from '@claude-flow/memory';

// Initialize vector storage with HNSW indexing
const adapter = new AgentDBAdapter({
  dimension: 1536,
  indexType: 'hnsw',
  metric: 'cosine',
  hnswM: 16,
  hnswEfConstruction: 200
});

// Store memory with embedding
await adapter.store({
  id: 'mem-123',
  content: 'User prefers dark mode',
  embedding: vector,
  metadata: { type: 'preference' }
});

// Semantic search (150x-12,500x faster than brute force)
const memories = await adapter.search(queryVector, {
  limit: 10,
  threshold: 0.7
});

// Cross-agent memory sharing
await adapter.enableCrossAgentSharing({
  shareTypes: ['patterns', 'preferences'],
  excludeTypes: ['secrets']
});

When to Use AgentDB

✅ **Memory & Pattern Storage**

    undefined

✅ **Semantic Search**

    undefined