Skip to main content
Cursor is an AI-powered code editor. You can integrate OpenMemoryX to give Cursor persistent memory across conversations.

​
Setup

​
1. Get Your API Key

  1. Log in to OpenMemoryX Portal
  2. Copy your default API key (starts with omx_)

​
2. Configure Cursor

In Cursor, you can use OpenMemoryX via:

​
Option A: Custom MCP Server

Create an MCP (Model Context Protocol) server that connects to OpenMemoryX: 
# mcp_server.py
from mcp.server import Server
import requests

app = Server("openmemoryx")

@app.call_tool()
def search_memories(query: str):
    """Search memories in OpenMemoryX"""
    response = requests.post(
        "https://t0ken.ai/api/v1/memories/search",
        headers={"X-API-Key": "omx_your_key"},
        json={"query": query, "limit": 5}
    )
    return response.json()["data"]

​
Option B: Direct API in Cursor Rules

Add to your .cursorrules file: 
## Memory Management

When working with this project, use OpenMemoryX for persistent memory:

- Store user preferences: POST https://t0ken.ai/api/v1/memories
- Retrieve context: POST https://t0ken.ai/api/v1/memories/search
- API Key: omx_your_key_here

Always check for relevant memories before making suggestions.

​
3. Using in Cursor

When chatting with Cursor, mention: 
"Remember that I prefer functional programming patterns"
"What did we decide about the database schema?"
"Search my memories for API design preferences"

​
Example Workflows

​
Store Code Preferences

User: I prefer using TypeScript with strict mode enabled
[Cursor stores: "User prefers TypeScript with strict mode"]

​
Retrieve Context

User: How should I structure this new API endpoint?
[Cursor searches memories for "API design preferences"]
[Cursor finds: "User prefers RESTful patterns with clear naming"]

​
Project-Specific Memory

User: This project uses PostgreSQL with Prisma
[Cursor stores with project_id: "current-project"]

​
Best Practices

  • Store decisions - Save architectural choices and preferences
  • Use project IDs - Separate memories per project
  • Regular cleanup - Delete outdated memories
  • Add metadata - Include context like tech stack