Skip to main content

​
Prerequisites

  • Cursor, Claude Code, or any MCP-compatible client
  • A QuantSpace account (for publishing and subscriptions)

​
1. Choose a server to connect

QuantX exposes multiple MCP servers. Add only the ones you need to your client config. 
{
  "mcpServers": {
    "quantx-marketplace": {
      "url": "https://quantx-marketplace.quantspace.io/mcp"
    },
    "quantx-signals": {
      "url": "https://quantx-signals.quantspace.io/mcp"
    }
  }
}
Restart your client. All tools from each connected server will appear automatically.

​
2. Discover a model

Ask your AI assistant:
“Search QuantX marketplace for momentum strategies on US equities”
The agent calls search_marketplace and returns a ranked list of available models. Each result includes a model_id.

​
3. Run a signal

Once you have a model_id, ask:
“Run the signal for model qx-mom-us-001 on AAPL, MSFT, NVDA from 2024-01-01 to 2025-01-01”
The agent calls run_signal on the Signals Server, waits for completion, and returns an output_url — ready to pass into a QuantLab portfolio optimization pipeline. 
search_marketplace → run_signal → [output_url] → run_po_job (QuantLab)

​
4. Server reference

ServerURLTools
Marketplacehttps://quantx-marketplace.quantspace.io/mcpsearch_marketplace, get_model, subscribe_to_model, publish_model
Signalshttps://quantx-signals.quantspace.io/mcprun_signal, get_signal_history, list_signals
Executionhttps://quantx-execution.quantspace.io/mcp(coming soon)