Risk_level in all 644 examples + model outputs risk classification

- All 644 examples tagged: 0=blocked(15), 1=refuse(33), 2=warn(24), 3=normal(498), 4=generous(74)
- Training output now includes risk_level field for decision transparency
- Model learns to classify risk before generating commands
- Validator can sanity-check: risk 0-1 should have empty commands

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

training/scripts/train_lora.py

@@ -79,8 +79,10 @@ def load_dataset(path: str) -> list:
 
             user_msg = "\n".join(user_parts)
 
-            # Assistant response as JSON
+            # Assistant response as JSON — includes risk_level for decision transparency
+            risk_level = ex.get("metadata", {}).get("risk_level", 3)
             response = {
+                "risk_level": risk_level,
                 "reasoning": out.get("reasoning", ""),
                 "commands": out.get("commands", []),
             }