Skip to content
Raindrop Developer Hub

SmartMemory App Deployment

Create and deploy a Raindrop application with SmartMemory that you can use with the MCP server. This gives you a deployed memory system that AI agents can access for persistent conversations and knowledge storage.

Prerequisites

  • Raindrop CLI installed and authenticated
  • Node.js 18+ installed
  • Basic understanding of AI agent memory concepts
  • Text editor of your choice

Create Your SmartMemory Project

Initialize a new Raindrop application:

Terminal window
raindrop build init my-smartmemory-app
cd my-smartmemory-app

Configure SmartMemory

Open the generated raindrop.manifest file and update it to include only SmartMemory:

raindrop.manifest
application "my-smartmemory-app" {
  smartmemory "agent-memory" {}
}

The SmartMemory resource will be deployed and accessible via the MCP server.

Generate Project Files

Run the generate command to create the necessary project structure:

Terminal window
raindrop build generate

This creates the project structure with your SmartMemory resource:

  • Directorymy-smartmemory-app/
    • raindrop.manifest
    • package.json
    • tsconfig.json

Deploy Your SmartMemory Resource

Deploy your SmartMemory resource to the Raindrop cloud:

Terminal window
raindrop build deploy --start

After deployment completes, you’ll see output similar to this:

Terminal window
Building to /Users/yourname/my-smartmemory-app/dist
Build successful
🔔 You deployed a full version, updates will require a full versioned deployment to work


📊 Watching deployment status...


my-smartmemory-app @01k0xyz...
Status: running


Modules (1)
  └─ agent-memory - running - no urls


──────────────────────────────────────────────────
Total: 1 modules (1 running)

Your SmartMemory resource is now deployed and ready to use with the MCP server.

Using with MCP Server

Your deployed SmartMemory resource can now be accessed through the LiquidMetal MCP server. The MCP server will connect to your deployed SmartMemory instance to manage AI agent memories across conversations.

When configuring the MCP server, you’ll reference your deployed SmartMemory application. The MCP server will then handle session management, memory storage, and retrieval for AI agents using your deployed infrastructure.

For complete setup instructions, see the Claude Code + Raindrop MCP Setup tutorial.

Understanding SmartMemory

Your deployed SmartMemory resource provides four types of memory for AI agents: working, episodic, semantic, and procedural memory. Each serves different purposes in giving agents persistent memory capabilities across conversations and sessions.

To learn more about how these memory types work together, see the SmartMemory concepts documentation.