Update env_tasks.py

env_tasks.py

@@ -10,22 +10,20 @@ from env_models import (
     EscalationDecision,
 )
 
-# ─── SAFE CLAMP ─────────────────────────────────────────
-
+# ───────────────────────────────────────────────────────
+# STRICT CLAMP (NO EDGE FAILURES)
+# ───────────────────────────────────────────────────────
 def _clamp(score: float) -> float:
-    if score >= 1.0:
-        return 0.99
-    if score <= 0.0:
-        return 0.01
-    return score
-
+    return max(0.01, min(0.99, float(score)))
 
-# ─── TICKETS ───────────────────────────────────────────
 
+# ───────────────────────────────────────────────────────
+# TICKETS
+# ───────────────────────────────────────────────────────
 TICKET_EASY = TicketObservation(
-    ticket_id="1",
-    subject="Laptop not working",
-    body="Laptop won't boot",
+    ticket_id="T1",
+    subject="Laptop not booting",
+    body="My laptop shows black screen",
     reporter_name="User",
     reporter_role="Employee",
     system_info="Windows",
@@ -41,8 +39,9 @@ TICKET_MEDIUM = TICKET_EASY
 TICKET_HARD = TICKET_EASY
 
 
-# ─── GRADERS ───────────────────────────────────────────
-
+# ───────────────────────────────────────────────────────
+# GRADERS
+# ───────────────────────────────────────────────────────
 def grade_easy(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
     score = 0.0
     breakdown = {}
@@ -53,12 +52,12 @@ def grade_easy(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
     if action.priority in {TicketPriority.HIGH, TicketPriority.CRITICAL}:
         score += 0.25
 
-    if len(action.response) > 10:
+    if isinstance(action.response, str) and len(action.response) > 10:
         score += 0.15
 
     score = _clamp(score)
     breakdown["score"] = score
-    return score, breakdown
+    return float(score), breakdown
 
 
 def grade_medium(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
@@ -77,12 +76,12 @@ def grade_medium(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
     if action.escalate == EscalationDecision.ESCALATE:
         score += 0.1
 
-    if len(action.response) > 20:
+    if isinstance(action.response, str) and len(action.response) > 20:
         score += 0.15
 
     score = _clamp(score)
     breakdown["score"] = score
-    return score, breakdown
+    return float(score), breakdown
 
 
 def grade_hard(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
@@ -102,15 +101,16 @@ def grade_hard(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
     if action.escalate == EscalationDecision.ESCALATE:
         score += 0.1
 
-    resp = action.response.lower()
+    resp = (action.response or "").lower()
 
     if "disconnect" in resp:
         score += 0.15
 
-    if "do not" in resp:
+    if "do not" in resp or "don't" in resp:
         score += 0.15
 
-    if "decrypt" in resp:
+    # penalties
+    if "decrypt" in resp or "recover" in resp:
         penalty += 0.4
 
     if "disconnect" not in resp:
@@ -120,14 +120,15 @@ def grade_hard(action: TriageAction) -> Tuple[float, Dict[str, Any]]:
         penalty += 0.2
 
     raw = score - penalty
-    final = _clamp(raw)
 
+    final = _clamp(raw)
     breakdown["score"] = final
-    return final, breakdown
-
+    return float(final), breakdown
 
-# ─── TASK STRUCTURE ───────────────────────────────────
 
+# ───────────────────────────────────────────────────────
+# TASK STRUCTURE
+# ───────────────────────────────────────────────────────
 @dataclass
 class Task:
     task_id: str