Skip to content

MCP (AI Assistants)

The Model Context Protocol (MCP) lets AI assistants like Cursor create strategies, run backtests, and manage your portfolio in natural language. MCP is separate from the REST API - it uses OAuth, not API keys.

https://api.quantly.trade/mcp

OpenAPI spec: quantly-mcp.yaml

1

In your MCP client (for example Cursor), add the Quantly MCP server URL above.

2

When prompted, sign in to Quantly and Authorize the connection. OAuth opens in your browser - approve access for the client you are connecting.

3

Ask your assistant to create strategies, backtest, or manage your portfolio in plain language.

Most MCP clients discover OAuth settings automatically. If your client needs manual configuration, the authorization server metadata is at https://api.quantly.trade/.well-known/oauth-authorization-server.

Through MCP, an authorized assistant can:

  • Create, update, copy, and delete strategies
  • Run backtests on saved or ad-hoc strategy definitions
  • Browse Explore strategies, Assets, and Track Record data
  • View portfolio, holdings, positions, and trade history
  • Execute portfolio actions (deposit, trade now, liquidate, skip today, and more) at your direction

MCP OAuth grants full trading API access to the connected client. It cannot change billing, delete your account, create API keys, or connect new broker accounts.

MCP REST API
Best for AI assistants (Cursor, Claude Desktop, etc.) Scripts, automation, custom integrations
Auth OAuth (browser sign-in) API keys (qk_live_...)
Setup Add MCP URL in your AI client Create a scoped key in Settings

For programmatic access from your own code, use the REST API with API keys instead.

  • Name the tickers or theme you care about (for example “momentum on tech giants” or “defensive when SPY RSI is above 70”).
  • Say whether you want a single stock, equal-weight basket, or a tilt toward one name.
  • Mention indicators or conditions when you have them (RSI, moving averages, return thresholds).
  • Start broad, then refine the generated tree in the Tree Builder.
  • In-app AI - click Generate with AI on the Strategies page. See AI Generation.
  • Manual build - use the Tree Builder directly.