MCP 服务器

AINative ZeroDB MCP Server

Provides AI assistants with vector search and persistent memory capabilities using ZeroDB. Enables storing and retrieving context across sessions, semantic document search, and analytics through natural language interactions.

README

🚀 AINative ZeroDB MCP Server

A Model Context Protocol (MCP) server that provides AI assistants with vector search and persistent memory capabilities using ZeroDB.

Features

🧠 Persistent Memory - Store and retrieve context across sessions
🔍 Vector Search - Semantic similarity search across documents
📊 Analytics - Query insights and usage patterns
🔐 Project Isolation - Separate data by project
⚡ High Performance - Optimized for real-time AI interactions

Quick Start

Installation

# Install globally
npm install -g ainative-zerodb-mcp-server

# Or use with npx
npx ainative-zerodb-mcp-server

Add to Claude Desktop

Add to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "zerodb": {
      "command": "npx",
      "args": ["-y", "ainative-zerodb-mcp-server"],
      "env": {
        "ZERODB_API_URL": "https://api.ainative.studio/api/v1",
        "ZERODB_PROJECT_ID": "your-project-id",
        "ZERODB_USERNAME": "your-username",
        "ZERODB_PASSWORD": "your-password"
      }
    }
  }
}

Add to Claude Code

# Add to current project
claude mcp add project zerodb npx -- -y ainative-zerodb-mcp-server

# Or add to .mcp.json

Available Tools

`zerodb_store_memory`

Store agent memory for persistent context across sessions.

{
  "content": "User prefers dark mode themes",
  "role": "assistant",
  "session_id": "session-123",
  "metadata": { "category": "preferences" }
}

`zerodb_retrieve_memory`

Retrieve stored memories and context.

{
  "session_id": "session-123",
  "limit": 10,
  "role": "assistant"
}

`zerodb_search`

Semantic search across stored documents and memories.

{
  "query": "user interface preferences",
  "limit": 5,
  "threshold": 0.7
}

`zerodb_store_vector`

Store embeddings for vector similarity search.

{
  "content": "Technical documentation about APIs",
  "vector": [0.1, 0.2, ...],
  "metadata": { "type": "documentation" }
}

`zerodb_analytics`

Query analytics and insights.

{
  "metric": "memory_usage",
  "period": "7d"
}

Configuration

Environment Variables

Variable	Description	Default
`ZERODB_API_URL`	ZeroDB API endpoint	`https://api.ainative.studio/api/v1`
`ZERODB_PROJECT_ID`	Your project identifier	Required
`ZERODB_USERNAME`	Authentication username	Required
`ZERODB_PASSWORD`	Authentication password	Required
`ZERODB_API_TOKEN`	API token (if available)	Optional
`MCP_CONTEXT_WINDOW`	Context window size	`8192`
`MCP_RETENTION_DAYS`	Data retention period	`30`

Usage Examples

With Claude Desktop

Once configured, you can use natural language:

"Remember that the user prefers dark mode"
"What do you know about my preferences?"
"Search for information about API documentation"
"Show me analytics for the last week"

Programmatic Usage

const { ZeroDBMCPServer } = require('ainative-zerodb-mcp-server');

const server = new ZeroDBMCPServer({
  apiUrl: 'https://api.ainative.studio/api/v1',
  projectId: 'my-project',
  username: 'user@example.com',
  password: 'secure-password'
});

await server.start();

Development

# Clone the repository
git clone https://github.com/AINative-Studio/ainative-zerodb-mcp-server.git
cd ainative-zerodb-mcp-server

# Install dependencies
npm install

# Run locally
npm start

# Run tests
npm test