Cursor IDE Integration¶

Integrating the COT Bias MCP Server with Cursor allows Cursor's AI (Chat, Composer, and Agent) to query real-time institutional positioning directly while you write code, backtest strategies, or conduct quantitative research.

⚡ Setup in 2 Steps¶

1. Configure `.cursor/mcp.json`¶

In the root of your workspace, create or edit .cursor/mcp.json:

{
  "mcpServers": {
    "cotbias": {
      "command": "uvx",
      "args": [
        "cotbias[mcp]",
        "cotbias-mcp"
      ],
      "env": {
        "COT_API_KEY": "cot_live_YOUR_ACTUAL_API_KEY"
      }
    }
  }
}

(Alternatively, configure globally via Cursor Settings → Features → MCP → Add New MCP Server).

2. Verify in Cursor Settings¶

Open Cursor Settings (Cmd + , on macOS, Ctrl + , on Windows).
Navigate to Features → MCP.
You will see cotbias listed with a green status indicator, showing 5 active tools.

💻 Workflow Examples in Cursor¶

1. Generating Backtest Code with COT Filters¶

In Cursor Composer (Cmd + I or Ctrl + I):

"Check the current COT positioning regime for E-mini S&P 500 (ES). Then write a Python script using pandas and backtrader that implements a trend-following strategy, using the COT 52-week index as a regime filter."

Cursor will call get_market_snapshot(symbol="ES"), inspect the regime, and generate Python code with accurate positioning parameters.

2. Scanning for Trading Setups¶

In Cursor Chat (Cmd + L):

"Scan the market universe for any contracts where momentum is 'New Longs (OI Expanding)' and crowdedness is 'Balanced'. Return the findings in a markdown table."

Cursor will call scan_markets and format the matched contracts directly in your chat.

3. Quantitative Timeseries Modeling¶

In Cursor Chat:

"Fetch 104 weeks of Gold (GC) timeseries data using cotbias and show me how to compute the correlation between commercial net positioning and next-month price returns."