tools/analytics_tool.py

from crewai_tools import RagTool
import json

class AnalyticsTool(RagTool):
    name: str = "analytics_tool"
    description: str = (
        "Processes and aggregates outputs from market, historical and sentiment agents"
         "to generate structured indicators and performance metrics, giving a holistic view of the cryptocurrency's condition."
    )

    def _run(self, market_data: dict, historical_data: dict, sentiment_data: dict):
        # combine results from the other agents into structured numeric metrics

        try:
            #Normalize strings -> dicts
            if isinstance(market_data, str):
                market_data = json.loads(market_data)
            if isinstance(historical_data, str):
                historical_data = json.loads(historical_data)
            if isinstance(sentiment_data, str):
                sentiment_data = json.loads(sentiment_data)

            #Extract info safely
            current_price = market_data.get("price") or market_data.get("latest_price")
            pct_change = historical_data.get("pct_change", 0)
            volatility = historical_data.get("volatility_pct", 0)
            trend = historical_data.get("trend", "unknown")
            sentiment = sentiment_data.get("final_sentiment_classification", "neutral")

            #Compute basic consistency logic
            aligned = (
                (trend == "upward" and "bullish" in sentiment.lower()) or
                (trend == "downward" and "bearish" in sentiment.lower())
            )

            score = (
                (pct_change/10) + (0.2 if aligned else -0.2)
                + (0.1 if "bullish" in sentiment.lower() else -0.1 if "bearish" in sentiment.lower() else 0)
            )

            score = round(max(-1, min(1, score)), 2)

            return {
                "price": current_price,
                "pct_change": pct_change,
                "volatility": volatility,
                "trend": trend,
                "sentiment": sentiment,
                "alignment": "aligned" if aligned else "divergent",
                "composite_score": score,
                "summary": f"Trend={trend}, Sentiment={sentiment}, Alignment={'aligned' if aligned else 'divergent'}, Score = {score}"
            }
        except Exception as e:
            return {"error": f"AnalyticsTool failed: {str(e)}"}