AI Assistant

Augy includes a built-in AI chat panel available in the agent editor and backtest results views. The assistant is context-aware -- it knows your current code, available indicators, the Agent API, and your backtest results.

• Requires an LLM API key configured in Settings (OpenAI or Anthropic).
• Responses stream in real time.
• API calls go directly from your machine to the LLM provider. No data passes through Augy servers.
• Chat history is saved per agent and per backtest for continuity across sessions.

When chatting from the agent editor, the AI has full context including:

• The complete Agent Python API documentation
• All 21 built-in indicator signatures and descriptions
• Your current agent code
• Any custom indicators you've created
• Recent backtest results for this agent (if available)

What you can ask

• "Help me write an RSI + MACD strategy with a trailing stop"
• "Why is my agent never entering trades?"
• "Add Bollinger Band squeeze detection to my strategy"
• "Explain what the Ichimoku cloud signals mean"

The AI responds with complete, copy-pasteable agent code so you can apply suggestions directly.

When chatting from a backtest results page, the AI has additional context:

• All performance metrics (return, Sharpe, drawdown, win rate, etc.)
• Backtest configuration (capital, fees, slippage, dataset)
• The agent code that produced the results

What you can ask

• "Why is my max drawdown so high?"
• "How can I improve the win rate without sacrificing returns?"
• "Is this Sharpe ratio good for a crypto strategy?"
• "Suggest improvements to reduce the number of losing trades"

Configure your API key in Settings. Only one provider needs to be configured.