| name | AgentDB Vector Search |
| description | Implement semantic vector search with AgentDB for intelligent document retrieval, similarity matching, and context-aware querying. Use when building RAG systems, semantic search engines, or intelligent knowledge bases. |
AgentDB Vector Search
What This Skill Does
Implements vector-based semantic search using AgentDB's high-performance vector database with 150x-12,500x faster operations than traditional solutions. Features HNSW indexing, quantization, and sub-millisecond search (<100µs).
Prerequisites
- Node.js 18+
- AgentDB v1.0.7+ (via agentic-flow or standalone)
- OpenAI API key (for embeddings) or custom embedding model
Quick Start with CLI
Initialize Vector Database
npx agentdb@latest init ./vectors.db
npx agentdb@latest init ./vectors.db --dimension 768
npx agentdb@latest init ./vectors.db --dimension 384
npx agentdb@latest init ./vectors.db --preset small
npx agentdb@latest init ./vectors.db --preset medium
npx agentdb@latest init ./vectors.db --preset large
npx agentdb@latest init ./vectors.db --in-memory
Query Vector Database
npx agentdb@latest query ./vectors.db "[0.1,0.2,0.3,...]"
npx agentdb@latest query ./vectors.db "[0.1,0.2,0.3]" -k 10
npx agentdb@latest query ./vectors.db "0.1 0.2 0.3" -t 0.75 -m cosine
npx agentdb@latest query ./vectors.db "[...]" -m euclidean
npx agentdb@latest query ./vectors.db "[...]" -m dot
npx agentdb@latest query ./vectors.db "[...]" -f json -k 5
npx agentdb@latest query ./vectors.db "[...]" -v
Import/Export Vectors
npx agentdb@latest export ./vectors.db ./backup.json
npx agentdb@latest import ./backup.json
npx agentdb@latest stats ./vectors.db
Quick Start with API
import { createAgentDBAdapter, computeEmbedding } from 'agentic-flow/reasoningbank';
const adapter = await createAgentDBAdapter({
dbPath: '.agentdb/vectors.db',
enableLearning: false,
enableReasoning: true,
quantizationType: 'binary',
cacheSize: 1000,
});
const text = "The quantum computer achieved 100 qubits";
const embedding = await computeEmbedding(text);
await adapter.insertPattern({
id: '',
type: 'document',
domain: 'technology',
pattern_data: JSON.stringify({
embedding,
text,
metadata: { category: "quantum", date: "2025-01-15" }
}),
confidence: 1.0,
usage_count: 0,
success_count: 0,
created_at: Date.now(),
last_used: Date.now(),
});
const queryEmbedding = await computeEmbedding("quantum computing advances");
const results = await adapter.retrieveWithReasoning(queryEmbedding, {
domain: 'technology',
k: 10,
useMMR: true,
synthesizeContext: true,
});
Core Features
1. Vector Storage
await db.storeWithEmbedding({
content: "Your document text",
metadata: { source: "docs", page: 42 }
});
2. Similarity Search
const similar = await db.findSimilar("quantum computing", {
limit: 5,
minScore: 0.75
});
3. Hybrid Search (Vector + Metadata)
const results = await db.hybridSearch({
query: "machine learning models",
filters: {
category: "research",
date: { $gte: "2024-01-01" }
},
limit: 20
});
Advanced Usage
RAG (Retrieval Augmented Generation)
async function ragQuery(question: string) {
const context = await db.searchSimilar(
await embed(question),
{ limit: 5, threshold: 0.7 }
);
const prompt = `Context: ${context.map(c => c.text).join('\n')}
Question: ${question}`;
return await llm.generate(prompt);
}
Batch Operations
await db.batchStore(documents.map(doc => ({
text: doc.content,
embedding: doc.vector,
metadata: doc.meta
})));
MCP Server Integration
npx agentdb@latest mcp
claude mcp add agentdb npx agentdb@latest mcp
Performance Benchmarks
npx agentdb@latest benchmark
Quantization Options
AgentDB provides multiple quantization strategies for memory efficiency:
Binary Quantization (32x reduction)
const adapter = await createAgentDBAdapter({
quantizationType: 'binary',
});
Scalar Quantization (4x reduction)
const adapter = await createAgentDBAdapter({
quantizationType: 'scalar',
});
Product Quantization (8-16x reduction)
const adapter = await createAgentDBAdapter({
quantizationType: 'product',
});
Distance Metrics
npx agentdb@latest query ./db.sqlite "[...]" -m cosine
npx agentdb@latest query ./db.sqlite "[...]" -m euclidean
npx agentdb@latest query ./db.sqlite "[...]" -m dot
Advanced Features
HNSW Indexing
- O(log n) search complexity
- Sub-millisecond retrieval (<100µs)
- Automatic index building
Caching
- 1000 pattern in-memory cache
- <1ms pattern retrieval
- Automatic cache invalidation
MMR (Maximal Marginal Relevance)
- Diverse result sets
- Avoid redundancy
- Balance relevance and diversity
Performance Tips
- Enable HNSW indexing: Automatic with AgentDB, 10-100x faster
- Use quantization: Binary (32x), Scalar (4x), Product (8-16x) memory reduction
- Batch operations: 500x faster for bulk inserts
- Match dimensions: 1536 (OpenAI), 768 (sentence-transformers), 384 (MiniLM)
- Similarity threshold: Start at 0.7 for quality, adjust based on use case
- Enable caching: 1000 pattern cache for frequent queries
Troubleshooting
Issue: Slow search performance
npx agentdb@latest stats ./vectors.db
Issue: High memory usage
Issue: Poor relevance
npx agentdb@latest query ./db.sqlite "[...]" -t 0.8
Issue: Wrong dimensions
npx agentdb@latest init ./db.sqlite --dimension 768
Database Statistics
npx agentdb@latest stats ./vectors.db
Performance Characteristics
- Vector Search: <100µs (HNSW indexing)
- Pattern Retrieval: <1ms (with cache)
- Batch Insert: 2ms for 100 vectors
- Memory Efficiency: 4-32x reduction with quantization
- Scalability: Handles 1M+ vectors efficiently
- Latency: Sub-millisecond for most operations
Learn More