Skip to main content

MCP Servers Overview

New to AINative?

See the Integration Checklist for the fast path: get a key, authenticate, and configure an MCP server in one linear sequence.

MCP Servers​

AINative publishes 7 MCP servers that give AI agents (Claude Code, Cursor, Windsurf, VS Code) direct access to your data layer, content system, design tokens, and analytics infrastructure — no REST calls required.

All Servers at a Glance​

ServerPackageToolsWhat It Does
ZeroDB Fullpip install zerodb-mcp69+Complete data layer: vectors, memory, tables, files, events, PostgreSQL, graph
ZeroDB Memorynpm i ainative-zerodb-memory-mcp10Agent memory + plan artifacts: store, recall, semantic search, versioned plans
PRD Generatornpm i ainative-prd-mcp18Generate, validate, and manage PRDs with AINative platform awareness + ZeroDB persistence
Sequential Thinkingnpx zerodb-sequential-thinking-mcp3Persistent reasoning chains with cross-session resume — self-provisioning
Design Systemnpm i -g ainative-design-system-mcp-server3Extract tokens, analyze components, generate themes
GTMnpx @ainative/gtm-mcp20Audit, fix, and publish Google Tag Manager containers
Strapi CMSnpm i -g ainative-strapi-mcp-server21Create, update, and publish Strapi content via natural language

Quick Setup (Auto-Detect)​

The fastest way to add ZeroDB to any editor:

npx zerodb-cli init

Auto-detects Claude Code, Cursor, VS Code, or Windsurf and writes the config.

Manual Setup — Claude Code​

Edit ~/.claude.json (global) or .claude/settings.json (project):

{
  "mcpServers": {
    "zerodb-memory": {
      "command": "npx",
      "args": ["-y", "ainative-zerodb-memory-mcp"],
      "env": {
        "ZERODB_API_KEY": "your-api-key",
        "ZERODB_PROJECT_ID": "your-project-id"
      }
    }
  }
}

Or use the CLI shorthand:

claude mcp add zerodb-memory npx -- -y ainative-zerodb-memory-mcp \
  -e ZERODB_API_KEY=your-key \
  -e ZERODB_PROJECT_ID=your-project-id

Manual Setup — Cursor​

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "zerodb": {
      "command": "python",
      "args": ["-m", "zerodb_mcp"],
      "env": {
        "ZERODB_API_KEY": "your-api-key",
        "ZERODB_PROJECT_ID": "your-project-id"
      }
    }
  }
}

Which Server Should I Use?​

I want to...Use
Give my agent persistent memoryZeroDB Memory
Store and version structured plans, PRDs, task listsZeroDB Memory (plan artifacts)
Multi-step reasoning that persists across sessionsSequential Thinking
Store vectors + files + tables + PostgreSQLZeroDB Full
Let an agent read/update my design tokens and component libraryDesign System
Automate Google Tag Manager — audit tags, fix misfires, publishGTM
Let an agent create/publish blog posts, tutorials, events in StrapiStrapi CMS
Run the server hosted (no local install)Remote MCP

Remote MCP (Hosted)​

Don't want to run any server locally? Use AINative's hosted remote MCP:

{
  "mcpServers": {
    "zerodb": {
      "url": "https://mcp.ainative.studio/zerodb",
      "headers": {
        "Authorization": "Bearer your-api-key"
      }
    }
  }
}

No install, no process management — the server runs in AINative's cloud.

REST Tool Discovery (No MCP Required)​

If your agent uses REST instead of MCP, you can discover all available ZeroDB tools via the Tool Catalog API:

## Full catalog
curl https://api.ainative.studio/api/v1/public/tools/catalog

## Filter by category
curl https://api.ainative.studio/api/v1/public/tools/catalog?category=memory

Returns machine-readable tool definitions with names, descriptions, REST endpoints, parameter schemas, and working examples. No authentication required.

The full OpenAPI spec is also available at /api/v1/openapi.json and interactive docs at /docs.

Security​

Before deploying agents that read private data or process untrusted content, read the Agent Security Guide.

It covers:

  • Scoped API keys — restrict each agent to only the permissions it needs, with TTL
  • Memory namespace isolation — use session:<id> namespaces for untrusted workflows
  • Content quarantine — strip prompt injection before passing content to agents
  • Heartbeat oversight — dead-man switch for long-running agents
  • Approval gates — human-in-the-loop for high-risk actions
  • Audit logs — full MCP tool call history with risk scores
tip

The Defense in Depth Checklist is a quick pre-deploy verification for any agent that touches real data.