Memory Layer for Cursor AI

Give your Cursor agent persistent memory across all coding sessions. Works natively as an MCP server — add to .cursor/mcp.json and your agent remembers context forever.

Installation

Cursor supports the Model Context Protocol (MCP) natively. Install the 0Latency MCP server:

npm install -g @0latency/mcp-server

Configuration

Add the MCP server to your Cursor config file at ~/.cursor/mcp.json:

{
  "mcpServers": {
    "0latency": {
      "command": "npx",
      "args": ["-y", "@0latency/mcp-server@latest"],
      "env": {
        "ZERO_LATENCY_API_KEY": "zl_your_api_key_here"
      }
    }
  }
}

Available Tools

Once configured, Cursor will have access to these memory tools:

• memory_add — Store memories from conversations
• memory_recall — Retrieve relevant context instantly
• memory_search — Semantic search across all memories
• memory_list — Browse and paginate stored memories
• memory_delete — Remove specific memories
• memory_graph — Explore entity relationships
• list_agents — Manage multiple agent namespaces

Usage Example

Once configured, your Cursor agent can automatically use memory tools:

You: Remember that I prefer using TypeScript strict mode
Cursor: [Uses memory_add to store this preference]

You: What language settings did I mention?
Cursor: [Uses memory_recall to retrieve your preferences]
         You mentioned preferring TypeScript strict mode.

Agent Namespaces

You can scope memories to different projects or contexts:

// Memories are automatically scoped to your API key
// Use different agent IDs for different projects:
// Set agent_id in memory operations to organize by project

Troubleshooting

MCP server not appearing in Cursor

Restart Cursor after modifying mcp.json. Check that Node.js and npm are installed and accessible.

Authentication errors

Verify your API key is correct and hasn't expired. Get a new key from your dashboard.