indicator validation looks for all NaN's and all zeroes

doc/plan.md

@@ -1,7 +1,5 @@
 # Development Plan
 
-* Wiki memory
-* Agent unification & spawn tool
 * Realtime data
 * Triggers
 * Strategy UI
@@ -13,6 +11,5 @@
 * Chat channels
   * MCP channel (with or without images)
 * TradingView indicator import tool
-* Trader preferences tool
-* Results persistence: research analysis, backtests, strategy performance metrics, etc.
-* 
+* Results persistence: ~~research analysis~~, backtests, strategy performance metrics, etc.
+* Free tier with token limits and sandbox shutdown

sandbox/dexorder/tools/indicator_harness.py

@@ -61,14 +61,33 @@ def make_synthetic_ohlcv(n: int = 200):
 def summarize(result, n: int) -> str:
     import pandas as pd
 
+    warnings = []
+
+    def _check_series_warnings(s: "pd.Series", label: str = "") -> None:
+        valid = s.dropna()
+        prefix = f"Column '{label}': " if label else ""
+        if len(valid) == 0:
+            warnings.append(
+                f"⚠ WARNING: {prefix}All {n} output values are NaN. "
+                "The indicator produced no usable data — check for NaN propagation bugs "
+                "(e.g. uninitialized recursive filter, division by zero, insufficient warmup)."
+            )
+        elif (valid == 0).all():
+            warnings.append(
+                f"⚠ WARNING: {prefix}All non-NaN output values are zero. "
+                "The indicator may have a computation bug "
+                "(e.g. log(1) always being 0, constant input, or a formula that cancels out)."
+            )
+
     if isinstance(result, pd.Series):
         nan_count = int(result.isna().sum())
         valid = result.dropna()
         sample = [round(float(v), 4) for v in valid.tail(5).values] if len(valid) else []
-        return (
+        summary = (
             f"Series({n} bars), NaN: {nan_count}/{n}, "
             f"last 5 valid values: {sample}"
         )
+        _check_series_warnings(result)
     elif isinstance(result, pd.DataFrame):
         cols = list(result.columns)
         nan_counts = {c: int(result[c].isna().sum()) for c in cols}
@@ -77,13 +96,18 @@ def summarize(result, n: int) -> str:
             valid = result[col].dropna()
             if len(valid):
                 sample[col] = [round(float(v), 4) for v in valid.tail(3).values]
-        return (
+            _check_series_warnings(result[col], label=col)
+        summary = (
             f"DataFrame({n} bars × {len(cols)} cols {cols}), "
             f"NaN counts: {nan_counts}, last 3 valid per col: {sample}"
         )
     else:
         return f"Unexpected return type: {type(result).__name__}"
 
+    if warnings:
+        return summary + "\n" + "\n".join(warnings)
+    return summary
+
 
 def run(impl_path: Path, metadata_path: Path) -> dict:
     """