AI‐Powered Market Indicator Plugin Generation

This guide explains how to use the curated AI prompt to generate market indicator plugins for the chart.py.

The prompt is tested and optimized for both ChatGPT and DeepSeek.

📋 Prerequisites

• Access to the charting application repository
• The AI prompt file: src/renderer/plugins/plugin-ai-prompt.md
• A ChatGPT or DeepSeek account

🚀 Quick Start

Step 1: Start a New AI Chat Session

Start a fresh conversation. For DeepSeek, choose Expert if necessary.

Step 2: Attach the Prompt File

Attach the entire src/renderer/plugins/plugin-ai-prompt.md file to the AI chat.

Or alternatively open the file and copy its entire contents into the prompt.

This prompt contains:

• Complete plugin system architecture documentation
• Function signature requirements
• Panel allocation rules
• Coding standards and helper utilities
• Output format specifications

The AI will acknowledge and ask which indicator you'd like to generate

Step 3: Request Your Indicator

Once the AI acknowledges the system prompt, provide the full name of the indicator you want. Alternatively you may add:

• specific default configuration,
• steps to calculate the indicator
• Any additional customization steps.

Example requests:

"Ichimoku Cloud"

"Money Flow Index (MFI) with a 14-period default and overbought/oversold lines at 80/20"

You can also request use of third party libraries like TA-Lib or pandas-ta. Make sure to install these libraries before running the plugins.

Step 4: Save the Generated Files

The AI will output exactly three sections plus notes:

1. Plugin File — Save as renderer/plugins/<module_name>.py
2. Configuration Block — Merge into defs/user.json under the CHART_PLUGINS object
3. Usage Example — The CLI command to activate the plugin
4. Notes — Important implementation details

Step 5: Enable and Test

1. Ensure the plugin module is saved in the correct location:

   renderer/plugins/<module_name>.py
   

2. Verify the configuration is properly merged into defs/user.json:

   {
     "CHART_PLUGINS": {
       "YOUR_INDICATOR_KEY": {
         "name": "module_name",
         "option": "cli-option-name",
         ...
       }
     }
   }

3. Run the chart with your new plugin:

   python chart.py --sym RELIANCE --cli-option-name

❗ Common Pitfalls

Plugin Not Loading

• Verify the "name" field in config matches the Python filename exactly (without .py)
• Check that the module is in renderer/plugins/
• Ensure no syntax errors in the generated code

Incorrect Slicing

• If you see too much or too little data, verify all make_addplot calls use .iloc[-display_period:]

Panel Conflicts

• If indicators overlap unexpectedly, check the panel configuration:
  • "share": false for exclusive panels
  • "allow_volume_panel": false to prevent sharing with volume
  • Adjust "preferred_axis" if two indicators conflict on the same y-axis

Missing Data

• The lookback value in config determines how much historical data is loaded. Increase it for indicators needing more history (e.g., 200-period MA needs "lookback": 200)