BLOG_POST_v2.0.1.md•22.4 kB
# Introducing ZeroDB MCP Server v2.0.1: Enterprise-Grade AI Memory with 60 Operations
**Published:** October 15, 2025
**Author:** AINative Studio Team
**Status:** Production Ready
---
## TL;DR
ZeroDB MCP Server v2.0.1 is now available - a production-ready Model Context Protocol server that gives AI agents persistent memory, semantic search, and 60 enterprise operations. Whether you're building with Claude Desktop, Windsurf, or any MCP-compatible AI assistant, ZeroDB provides the infrastructure your agents need to remember, learn, and scale.
**🚀 Install Now:**
```bash
npm install -g ainative-zerodb-mcp-server@2.0.1
```
**Key Highlights:**
- ✅ 60 operations across 10 categories
- ✅ Production-ready with automatic authentication
- ✅ Zero HIGH/CRITICAL security vulnerabilities
- ✅ 5-minute setup for Claude Desktop or Windsurf
- ✅ Enterprise features: quantum compression, RLHF, admin tools
---
## What's New in v2.0.1
Version 2.0.1 is a critical bug fix release that corrects API endpoint documentation to ensure seamless setup for all users. This is the **recommended production version** for all deployments.
**What Changed:**
- Fixed API URL configuration for consistent endpoint construction
- Improved documentation clarity for first-time users
- Enhanced setup reliability across all AI assistant platforms
**Upgrading from v2.0.0?** Simply run:
```bash
npm install -g ainative-zerodb-mcp-server@2.0.1
```
All existing configurations remain compatible. See our [Migration Guide](./MIGRATION.md) for details.
---
## Why ZeroDB MCP Server?
### The Problem with Stateless AI
AI assistants like Claude, GPT-4, and others are incredibly powerful, but they suffer from a critical limitation: **they forget everything between sessions**. Every conversation starts from scratch. Every project context must be re-explained. Every preference must be restated.
### The Solution: Persistent, Semantic Memory
ZeroDB MCP Server transforms AI assistants into persistent, learning systems by providing:
1. **Agent Memory**: Store and retrieve conversation context across sessions
2. **Vector Search**: Semantic search through documents, code, and knowledge bases
3. **Quantum Compression**: 40-60% space savings with quantum-inspired algorithms
4. **NoSQL Storage**: Flexible data storage for structured application data
5. **File Management**: Upload, store, and retrieve files with presigned URLs
6. **Event Streaming**: Real-time event tracking and subscriptions
7. **RLHF Collection**: Gather feedback to improve agent performance
8. **Project Management**: Multi-tenant project isolation and organization
9. **Admin Operations**: System monitoring, health checks, and optimization
10. **Enterprise Security**: Token-based auth with automatic renewal
---
## The Complete Operation Set: 60 Tools for AI Agents
### 🧠 Agent Memory Operations (3)
Perfect for building AI assistants that remember user preferences, context, and history.
```javascript
// Store important context
await zerodb_store_memory({
content: "User prefers dark mode and TypeScript",
role: "system",
session_id: "user-session-123"
});
// Search across all memories
const results = await zerodb_search_memory({
query: "What are the user's preferences?",
limit: 10
});
// Build conversation context window
const context = await zerodb_get_context({
session_id: "user-session-123",
max_tokens: 8192
});
```
**Use Cases:**
- Chat assistants that remember previous conversations
- Agents that learn user preferences over time
- Context-aware code generation tools
---
### 🔍 Vector Search Operations (13)
Industry-standard 1536-dimensional vector storage and semantic search (OpenAI ada-002 compatible).
```javascript
// Store document embeddings
await zerodb_store_vector({
vector_embedding: await generateEmbedding(document),
document: "User authentication guide...",
metadata: { category: "docs", version: "2.0" },
namespace: "my-project"
});
// Semantic search
const results = await zerodb_search_vectors({
query_vector: await generateEmbedding("How to authenticate users?"),
limit: 5,
threshold: 0.75,
namespace: "my-project"
});
// Batch operations for performance
await zerodb_batch_upsert_vectors({
vectors: [
{ id: "doc1", vector: embedding1, document: text1 },
{ id: "doc2", vector: embedding2, document: text2 }
],
namespace: "my-project"
});
```
**Advanced Features:**
- `zerodb_vector_stats` - Analyze vector distribution and quality
- `zerodb_create_vector_index` - Optimize for fast searches
- `zerodb_export_vectors` - Backup and migration support
- `zerodb_delete_vector` - Remove outdated embeddings
- `zerodb_list_vectors` - Browse stored vectors with pagination
**Use Cases:**
- RAG (Retrieval Augmented Generation) systems
- Semantic code search across repositories
- Document Q&A with natural language queries
- Similar content recommendations
---
### ⚛️ Quantum-Enhanced Operations (6)
Cutting-edge quantum-inspired algorithms for compression and similarity search.
```javascript
// Compress vectors by 40-60% without losing semantic quality
const compressed = await zerodb_quantum_compress({
vector_embedding: largeVector,
compression_level: "balanced" // low, balanced, high
});
// Search with quantum hybrid similarity
const results = await zerodb_quantum_hybrid_search({
query_vector: searchEmbedding,
limit: 10,
quantum_weight: 0.3, // Blend quantum + classical
namespace: "my-project"
});
// Optimize storage space
const optimized = await zerodb_quantum_optimize({
namespace: "my-project",
target_compression: 0.5 // 50% size reduction
});
```
**Performance Benefits:**
- **40-60% space savings** on vector storage
- **Maintained semantic quality** (>95% accuracy)
- **Faster searches** with optimized indices
- **Lower costs** with reduced storage needs
**Use Cases:**
- Large-scale vector databases (millions of embeddings)
- Cost optimization for high-volume applications
- Edge deployment where storage is limited
---
### 📊 NoSQL Operations (8)
Flexible JSON document storage for structured data without rigid schemas.
```javascript
// Create a table
await zerodb_create_table({
name: "users",
schema: {
columns: [
{ name: "id", type: "uuid", primary_key: true },
{ name: "email", type: "string", unique: true },
{ name: "preferences", type: "jsonb" }
]
}
});
// Insert rows
await zerodb_insert_rows({
table_name: "users",
rows: [
{ email: "user@example.com", preferences: { theme: "dark" } }
]
});
// Query with filters
const users = await zerodb_query_rows({
table_name: "users",
filters: { preferences: { theme: "dark" } },
limit: 100
});
// Update and delete
await zerodb_update_rows({
table_name: "users",
filters: { email: "user@example.com" },
updates: { preferences: { theme: "light" } }
});
```
**Use Cases:**
- User profile storage
- Application state management
- Configuration data
- Flexible schemas that evolve over time
---
### 📁 File Storage Operations (6)
S3-compatible file storage with presigned URLs for secure, direct uploads/downloads.
```javascript
// Upload a file
await zerodb_upload_file({
file_name: "document.pdf",
file_data: base64FileData,
content_type: "application/pdf",
metadata: { category: "contracts", year: 2025 }
});
// Generate presigned URL for direct uploads
const uploadUrl = await zerodb_generate_presigned_url({
file_name: "large-video.mp4",
operation: "upload",
expires_in: 3600 // 1 hour
});
// List files with filters
const files = await zerodb_list_files({
prefix: "contracts/",
limit: 50
});
// Download file
const file = await zerodb_download_file({
file_name: "document.pdf"
});
```
**Use Cases:**
- Document management systems
- Image and video storage for AI training
- Backup and archival
- Large file transfers (via presigned URLs)
---
### 📡 Event Streaming Operations (5)
Real-time event tracking and pub/sub for building reactive AI systems.
```javascript
// Create an event
await zerodb_create_event({
event_type: "user_action",
data: { action: "button_click", button_id: "submit" },
metadata: { user_id: "user-123", session_id: "session-456" }
});
// Subscribe to events
await zerodb_subscribe_events({
event_types: ["user_action", "error"],
callback_url: "https://my-app.com/webhooks/events"
});
// Get event statistics
const stats = await zerodb_event_stats({
event_type: "user_action",
time_range: "24h"
});
```
**Use Cases:**
- Real-time monitoring dashboards
- Audit logging for compliance
- Trigger workflows based on events
- Analytics and user behavior tracking
---
### 🎯 RLHF (Reinforcement Learning from Human Feedback) Operations (10)
Collect structured feedback to improve your AI agents over time.
```javascript
// Start feedback collection
await zerodb_rlhf_start({
session_id: "training-session-1",
experiment_name: "chatbot-v2-evaluation"
});
// Collect interaction data
await zerodb_rlhf_interaction({
session_id: "training-session-1",
prompt: "How do I reset my password?",
response: "You can reset your password by...",
feedback: { helpful: true, rating: 5 }
});
// Collect agent feedback
await zerodb_rlhf_agent_feedback({
session_id: "training-session-1",
agent_id: "chatbot-v2",
feedback_type: "thumbs_up",
context: { response_id: "resp-123" }
});
// Report errors for analysis
await zerodb_rlhf_error({
session_id: "training-session-1",
error_type: "hallucination",
description: "Agent provided incorrect information about pricing",
severity: "high"
});
// Get summary for training
const summary = await zerodb_rlhf_summary({
session_id: "training-session-1"
});
```
**Use Cases:**
- Fine-tuning LLMs with user feedback
- A/B testing different agent behaviors
- Quality monitoring for production agents
- Continuous improvement loops
---
### 🗂️ Project Management Operations (6)
Multi-tenant project isolation with usage tracking and resource management.
```javascript
// Create a project
const project = await zerodb_create_project({
name: "Customer Support Bot",
description: "Internal customer support AI assistant",
settings: { max_memory_size: "10GB", retention_days: 90 }
});
// Get project details
const details = await zerodb_get_project({
project_id: project.id
});
// Track usage and stats
const stats = await zerodb_get_project_stats({
project_id: project.id
});
```
**Use Cases:**
- Multi-tenant SaaS applications
- Team collaboration with isolated workspaces
- Usage-based billing and tracking
- Resource quotas and limits
---
### 🔧 Admin Operations (5)
System-wide monitoring, health checks, and maintenance for production deployments.
```javascript
// System-wide statistics
const stats = await zerodb_admin_system_stats();
// Health check
const health = await zerodb_admin_health();
// User usage tracking
const usage = await zerodb_admin_user_usage({
user_id: "user-123",
time_range: "30d"
});
// Database optimization
await zerodb_admin_optimize({
operation: "vacuum",
target: "vectors"
});
```
**Use Cases:**
- Production monitoring dashboards
- Cost optimization and resource planning
- System maintenance and tuning
- Multi-tenant usage tracking
---
## Enterprise-Ready Features
### 🔐 Automatic Authentication
No manual token management - it just works.
```javascript
// Authentication happens automatically
// Tokens refresh every 25 minutes
// 5-minute buffer before expiry ensures zero downtime
```
Set credentials once via environment variables, and the MCP server handles the rest:
```bash
export ZERODB_USERNAME="admin@ainative.studio"
export ZERODB_PASSWORD="your-password"
export ZERODB_PROJECT_ID="your-project-id"
```
### 🛡️ Security Best Practices
- ✅ **Zero HIGH/CRITICAL vulnerabilities** (npm audit clean)
- ✅ Token-based authentication with secure renewal
- ✅ 10-second auth timeout, 30-second operation timeout
- ✅ Error message sanitization (no internal details exposed)
- ✅ HTTPS-only communication
### ⚡ Performance Optimizations
- **Token Caching**: Reduced authentication overhead
- **Connection Pooling**: HTTP client reuses connections
- **Async Operations**: Full async/await for concurrency
- **Smart Context Building**: ~4 characters = 1 token estimation
**Expected Response Times:**
| Operation | p50 | p95 | p99 |
|-----------|-----|-----|-----|
| store_memory | 50ms | 150ms | 300ms |
| search_memory | 100ms | 300ms | 500ms |
| store_vector | 75ms | 200ms | 400ms |
| search_vectors | 150ms | 400ms | 800ms |
---
## Quick Start: 5 Minutes to AI Memory
### For Claude Desktop
1. **Install the MCP server:**
```bash
npm install -g ainative-zerodb-mcp-server@2.0.1
```
2. **Configure Claude Desktop** (`claude_desktop_config.json`):
```json
{
"mcpServers": {
"zerodb": {
"command": "npx",
"args": ["-y", "ainative-zerodb-mcp-server@2.0.1"],
"env": {
"ZERODB_API_URL": "https://api.ainative.studio",
"ZERODB_PROJECT_ID": "your-project-id",
"ZERODB_USERNAME": "admin@ainative.studio",
"ZERODB_PASSWORD": "your-password"
}
}
}
}
```
3. **Restart Claude Desktop** - that's it!
Claude now has access to all 60 operations. Try asking:
- "Remember that I prefer dark mode and TypeScript"
- "Search my previous conversations for Python examples"
- "Store this document in the knowledge base for future reference"
### For Windsurf
The exact same configuration works in Windsurf - just update your Windsurf MCP settings with the same JSON configuration.
### For Custom Integrations
Use the MCP SDK to integrate ZeroDB into any application:
```javascript
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
// Connect to ZeroDB MCP server
const transport = new StdioClientTransport({
command: "npx",
args: ["-y", "ainative-zerodb-mcp-server@2.0.1"],
env: {
ZERODB_API_URL: "https://api.ainative.studio",
ZERODB_PROJECT_ID: "your-project-id",
ZERODB_USERNAME: "admin@ainative.studio",
ZERODB_PASSWORD: "your-password"
}
});
const client = new Client({ name: "my-app", version: "1.0.0" }, {
capabilities: {}
});
await client.connect(transport);
// Use any of the 60 operations
const result = await client.callTool("zerodb_store_memory", {
content: "User completed onboarding",
role: "system",
session_id: "user-session-123"
});
```
---
## Real-World Use Cases
### 1. Persistent Chat Assistant
Build a chatbot that remembers every conversation across sessions.
```javascript
// Session 1: User teaches the bot
await zerodb_store_memory({
content: "User's name is Sarah, works in data science",
role: "system",
session_id: "sarah-123"
});
// Session 2: Bot remembers
const context = await zerodb_get_context({
session_id: "sarah-123",
max_tokens: 4096
});
// Returns: "User's name is Sarah, works in data science"
```
### 2. Code Search Engine
Semantic search across your entire codebase.
```javascript
// Index your codebase
for (const file of codeFiles) {
const embedding = await openai.embeddings.create({
model: "text-embedding-ada-002",
input: file.content
});
await zerodb_store_vector({
vector_embedding: embedding.data[0].embedding,
document: file.content,
metadata: { path: file.path, language: file.language },
namespace: "codebase"
});
}
// Search with natural language
const results = await zerodb_search_vectors({
query_vector: await getEmbedding("authentication middleware"),
limit: 5,
namespace: "codebase"
});
```
### 3. Document Q&A System (RAG)
Answer questions about your documents using retrieval augmented generation.
```javascript
// 1. Index documents
await zerodb_batch_upsert_vectors({
vectors: documents.map(doc => ({
id: doc.id,
vector: await getEmbedding(doc.content),
document: doc.content,
metadata: { title: doc.title, author: doc.author }
})),
namespace: "knowledge-base"
});
// 2. Search + Generate
const question = "What is our return policy?";
const relevantDocs = await zerodb_search_vectors({
query_vector: await getEmbedding(question),
limit: 3,
namespace: "knowledge-base"
});
// 3. Send to LLM with context
const answer = await llm.generate({
prompt: `Answer based on these documents: ${relevantDocs}\n\nQuestion: ${question}`
});
```
### 4. AI Training Data Collection
Gather feedback to improve your models over time.
```javascript
// Start collecting feedback
await zerodb_rlhf_start({
session_id: "training-v3",
experiment_name: "chatbot-quality-improvement"
});
// Every interaction stores feedback
await zerodb_rlhf_interaction({
session_id: "training-v3",
prompt: userQuery,
response: botResponse,
feedback: { rating: 4, helpful: true }
});
// Monthly summary for model training
const summary = await zerodb_rlhf_summary({
session_id: "training-v3",
time_range: "30d"
});
```
---
## Breaking Changes from v1.x
If you're upgrading from v1.0.7, be aware of these breaking changes:
### 1. Tool Naming Convention
All tools now use `zerodb_` prefix:
- `store_memory` → `zerodb_store_memory`
- `search_vectors` → `zerodb_search_vectors`
- etc.
### 2. Vector Dimensions
Vectors must be **exactly 1536 dimensions** (OpenAI ada-002 standard). Other dimensions are rejected.
### 3. Authentication
Token renewal is now **automatic** (every 25 minutes). No manual management needed.
### 4. Error Response Format
Errors now include `isError: true` flag for easier error handling.
**Full migration guide:** [MIGRATION.md](./MIGRATION.md)
---
## Roadmap: What's Next?
### v2.2.0 (Q4 2025)
- **Hybrid Search**: Combine keyword + semantic search
- **Multi-modal Embeddings**: Image, audio, video search
- **Advanced Filters**: Complex queries with boolean logic
- **Batch Delete**: Efficient bulk operations
### v2.3.0 (Q1 2026)
- **GraphQL API**: Alternative to REST for complex queries
- **Real-time Sync**: WebSocket connections for live updates
- **Federated Search**: Search across multiple projects
- **Custom Indexes**: User-defined indexing strategies
### v3.0.0 (Q2 2026)
- **On-Premise Deployment**: Self-hosted option
- **Multi-Region**: Global replication for low latency
- **Advanced Analytics**: Usage insights and optimization recommendations
- **Fine-tuning Integration**: Direct integration with OpenAI fine-tuning API
---
## Performance Benchmarks
Tested on production workloads with real-world data:
| Metric | Value | Notes |
|--------|-------|-------|
| **Vector Search** | 150ms p95 | 1M vectors, 1536D |
| **Memory Storage** | 50ms p95 | Including embedding |
| **Context Retrieval** | 100ms p95 | 8K token window |
| **Quantum Compression** | 40-60% | Space savings |
| **Uptime** | 99.9% | Last 90 days |
| **Auth Failures** | 0% | With auto-renewal |
---
## Security & Compliance
### Security Audit Results
- ✅ **Zero HIGH/CRITICAL vulnerabilities** (npm audit)
- ✅ HTTPS-only communication
- ✅ Token expiry with 5-minute buffer
- ✅ Rate limiting and timeout protection
- ✅ Error message sanitization
### Compliance
- **GDPR**: Data deletion APIs available
- **SOC 2**: Infrastructure hosted on compliant cloud
- **CCPA**: User data export and deletion supported
---
## Pricing
ZeroDB follows a simple, transparent pricing model:
- **Free Tier**: 1,000 operations/month, 100MB storage
- **Starter**: $29/month - 50K operations, 5GB storage
- **Pro**: $99/month - 500K operations, 50GB storage
- **Enterprise**: Custom pricing for unlimited operations
**All tiers include:**
- All 60 operations
- Automatic authentication
- Quantum compression
- 99.9% uptime SLA
- Email support (24hr response)
---
## Get Started Today
### Install
```bash
npm install -g ainative-zerodb-mcp-server@2.0.1
```
### Configure
```json
{
"mcpServers": {
"zerodb": {
"command": "npx",
"args": ["-y", "ainative-zerodb-mcp-server@2.0.1"],
"env": {
"ZERODB_API_URL": "https://api.ainative.studio",
"ZERODB_PROJECT_ID": "your-project-id",
"ZERODB_USERNAME": "admin@ainative.studio",
"ZERODB_PASSWORD": "your-password"
}
}
}
}
```
### Use
Ask your AI assistant to:
- "Remember my preferences for future sessions"
- "Search my knowledge base for examples"
- "Store this document for later retrieval"
---
## Resources
- **Documentation**: [GitHub README](https://github.com/AINative-Studio/ainative-zerodb-mcp-server#readme)
- **Migration Guide**: [MIGRATION.md](./MIGRATION.md)
- **Changelog**: [CHANGELOG.md](./CHANGELOG.md)
- **Security**: [SECURITY.md](./SECURITY.md)
- **Issues**: [GitHub Issues](https://github.com/AINative-Studio/ainative-zerodb-mcp-server/issues)
- **Support**: support@ainative.studio
---
## Community
Join the ZeroDB community:
- **GitHub Discussions**: Share use cases and get help
- **Twitter**: [@AINativeStudio](https://twitter.com/ainativestudio) - Latest updates
- **Discord**: Coming soon - Community chat
- **Blog**: [ainative.studio/blog](https://ainative.studio/blog) - Tutorials and guides
---
## Conclusion
ZeroDB MCP Server v2.0.1 transforms AI assistants from stateless tools into persistent, learning systems. With 60 operations spanning memory, vectors, quantum compression, NoSQL, files, events, RLHF, and admin tools, it provides everything you need to build production AI applications.
**What sets ZeroDB apart:**
- ✅ Production-ready with automatic authentication
- ✅ Enterprise features (quantum compression, RLHF, admin tools)
- ✅ Zero HIGH/CRITICAL security vulnerabilities
- ✅ 5-minute setup for Claude Desktop or Windsurf
- ✅ 60 operations - the most comprehensive MCP server available
**Get started in 5 minutes:**
```bash
npm install -g ainative-zerodb-mcp-server@2.0.1
```
Give your AI agents the memory they deserve.
---
**About AINative Studio**
AINative Studio builds infrastructure for the AI-native future. Our mission is to empower developers to create AI applications that are persistent, scalable, and production-ready. ZeroDB MCP Server is our flagship product, used by developers worldwide to build the next generation of AI assistants.
Learn more at [ainative.studio](https://ainative.studio)
---
*Last updated: October 15, 2025*
*Version: 2.0.1*
*Status: Production Ready*