ricable RuVector
High-performance vector database for Node.js with automatic native Rust NAPI / WASM fallback. Provides the unified CLI entry point for the entire RuVector ecosystem.
Quick Command Reference
| Task | Command |
|---|---|
| Show help | npx ruvector@latest --help |
| Create database | npx ruvector@latest create --dimensions 384 |
| Insert vectors | npx ruvector@latest insert --file vectors.json |
| Search | npx ruvector@latest search --query "hello world" --top-k 10 |
| Build HNSW index | npx ruvector@latest index build --ef-construction 200 |
| Benchmark | npx ruvector@latest bench --dimensions 384 --count 10000 |
| Show info | npx ruvector@latest info |
| Start server | npx ruvector@latest serve --port 8080 |
Installation
Hub install (recommended): npx ruvector@latest is the main entry point for the full ecosystem.
Standalone: npx ruvector@latest
See Installation Guide for the full ecosystem.
Core Commands
Database Operations
npx ruvector@latest create --dimensions 384 --metric cosine # Create new DB
npx ruvector@latest insert --file vectors.json # Bulk insert
npx ruvector@latest insert --id vec-1 --vector "[0.1, 0.2]" # Single insert
npx ruvector@latest search --query "[0.1, 0.2]" --top-k 10 # Vector search
npx ruvector@latest delete --id vec-1 # Delete vector
npx ruvector@latest count # Count vectors
npx ruvector@latest info # DB information
Index Management
npx ruvector@latest index build --ef-construction 200 --m 16 # Build HNSW
npx ruvector@latest index status # Index status
npx ruvector@latest index optimize # Optimize index
npx ruvector@latest index rebuild # Rebuild index
Server Mode
npx ruvector@latest serve --port 8080 # Start HTTP server
npx ruvector@latest serve --port 8080 --grpc 50051 # HTTP + gRPC
npx ruvector@latest serve --tls --cert cert.pem --key key.pem # With TLS
Benchmarking
npx ruvector@latest bench --dimensions 384 --count 10000 # Insert benchmark
npx ruvector@latest bench --mode search --queries 1000 # Search benchmark
npx ruvector@latest bench --mode mixed --duration 60 # Mixed workload
Programmatic API
import { VectorDB } from 'ruvector';
const db = new VectorDB({ dimensions: 384, metric: 'cosine' });
// Insert vectors
await db.insert('vec-1', [0.1, 0.2, ...], { label: 'example' });
// Batch insert
await db.batchInsert(vectors); // 50k+ inserts/sec
// Search
const results = await db.search(queryVector, { topK: 10 });
// Build HNSW index
await db.buildIndex({ efConstruction: 200, m: 16 });
Key Options:
| Option | Description | Default |
|---|---|---|
--dimensions |
Vector dimensionality | Required |
--metric |
Distance metric: cosine, euclidean, dot |
cosine |
--ef-construction |
HNSW construction parameter | 200 |
--m |
HNSW max connections per layer | 16 |
--top-k |
Number of results to return | 10 |
Common Patterns
Embed and Search Workflow
# Create database, insert embeddings, and search
npx ruvector@latest create --dimensions 384 --metric cosine
npx ruvector@latest insert --file embeddings.json
npx ruvector@latest index build
npx ruvector@latest search --query "[0.1, 0.2, ...]" --top-k 5
Production Server Deployment
# Start with persistence and authentication
npx ruvector@latest serve --port 8080 --persist ./data --auth-token $RV_TOKEN
Benchmark Before Tuning
# Run full benchmark suite
npx ruvector@latest bench --dimensions 384 --count 50000 --mode mixed
RAN DDD Context
Bounded Context: Data Infrastructure
References
- Command reference: See references/commands.md
- Full README
- npm