Weaviate

Weaviate is an open-source vector database with support for cloud and self-hosted deployments. The TONL-MCP Bridge Weaviate adapter provides seamless integration with TONL conversion.

Installation

npm install tonl-mcp-bridge weaviate-client

Basic Usage

import { WeaviateAdapter } from 'tonl-mcp-bridge';

// Cloud deployment
const adapter = new WeaviateAdapter({
  url: 'https://your-cluster.weaviate.network',
  apiKey: process.env.WEAVIATE_API_KEY
});

// Local deployment
const adapter = new WeaviateAdapter({
  host: 'localhost',
  scheme: 'http'
});

await adapter.connect();

// Search and convert to TONL
const results = await adapter.search('Article', embedding, {
  limit: 10
});

// With statistics
const { tonl, stats } = await adapter.searchWithStats('Article', embedding, {
  limit: 10,
  model: 'gpt-4o'
});

console.log(`Saved ${stats.savingsPercent}% tokens`);

await adapter.disconnect();

Configuration Options

// Weaviate Cloud
const adapter = new WeaviateAdapter({
  url: 'https://cluster.weaviate.network',
  apiKey: 'your-api-key'
});

// Self-hosted
const adapter = new WeaviateAdapter({
  host: 'localhost',
  scheme: 'http'  // or 'https'
});

// Custom host and port
const adapter = new WeaviateAdapter({
  host: 'weaviate.example.com',
  scheme: 'https'
});

Search Options

await adapter.search('Article', embedding, {
  limit: 10,
  minScore: 0.7  // Minimum similarity score
});

Weaviate Setup

1. Choose Deployment:

   • Weaviate Cloud - Fully managed
   • Docker - Self-hosted
   • Kubernetes - Production

2. Create Schema:

# Example: Create Article class
client.schema.create_class({
    "class": "Article",
    "vectorizer": "text2vec-openai",  # or other vectorizer
    "properties": [
        {
            "name": "title",
            "dataType": ["text"]
        },
        {
            "name": "content",
            "dataType": ["text"]
        },
        {
            "name": "category",
            "dataType": ["text"]
        }
    ]
})

3. Import Data:

# Use Weaviate client for imports
client.batch.configure(batch_size=100)
with client.batch as batch:
    for obj in objects:
        batch.add_data_object(
            data_object=obj,
            class_name="Article"
        )

# Then search with TONL adapter
const adapter = new WeaviateAdapter({...});
const results = await adapter.searchToTonl('Article', queryEmbedding);

Weaviate Features

• Automatic Vectorization: Built-in text2vec modules
• Hybrid Search: Combine vector and keyword search
• Multi-tenancy: Isolated data per tenant
• Generative Search: Built-in LLM integration
• GraphQL API: Flexible querying

Docker Quick Start

# Start Weaviate locally
docker run -d \
  -p 8080:8080 \
  -e AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED=true \
  semitechnologies/weaviate:latest

# Use with TONL adapter
const adapter = new WeaviateAdapter({
  host: 'localhost',
  scheme: 'http'
});

Best Practices

1. Choose Right Vectorizer for your data type
2. Use Multi-tenancy for SaaS applications
3. Configure Replication for high availability
4. Monitor Resource Usage with Weaviate metrics

Weaviate vs MongoDB

Feature	Weaviate	MongoDB
Vectorization	Built-in	Manual
Hybrid Search	Native	Native
Setup Complexity	Higher	Lower
Nested Objects	No	Yes (60% savings)
Already Installed	No	Yes (millions)

Next Steps

• Weaviate Documentation - Official docs
• MongoDB Guide - Compare alternatives
• ROI Calculator - Calculate savings