feat+test: add drift detection in record_decision and 57 unit tests for FrameTracer/DivergenceDetector

agent_debugger_sdk/core/context/trace_context.py

@@ -188,6 +188,8 @@ async def restore(
         ctx._restored_state = restored_state
         ctx.replayed_events: list[dict[str, Any]] = []
         ctx._drift_detector = None
+        ctx._drift_events: list[Any] = []
+        ctx._drift_compare_index = 0
         ctx._hook_errors: list[Exception] = []
         ctx._restored_target: Any = None
 

agent_debugger_sdk/core/recorders.py

@@ -98,8 +98,9 @@ async def record_decision(
         self,
         reasoning: str,
         confidence: float,
-        evidence: list[dict[str, Any]],
         chosen_action: str,
+        *,
+        evidence: list[dict[str, Any]] | None = None,
         evidence_event_ids: list[str] | None = None,
         upstream_event_ids: list[str] | None = None,
         alternatives: list[dict[str, Any]] | None = None,
@@ -114,14 +115,42 @@ async def record_decision(
             name=name,
             reasoning=reasoning,
             confidence=max(0.0, min(1.0, confidence)),
-            evidence=evidence,
+            evidence=evidence or [],
             evidence_event_ids=evidence_event_ids or [],
             alternatives=alternatives or [],
             chosen_action=chosen_action,
             importance=0.7,
             upstream_event_ids=upstream_event_ids or [],
         )
         await self._emit_event(event)
+
+        # Detect drift against the original execution if a detector is active
+        drift_detector = getattr(self, "_drift_detector", None)
+        if drift_detector is not None:
+            drift_index = getattr(self, "_drift_compare_index", 0)
+            original_events = getattr(drift_detector, "original_events", [])
+            # Advance to the current decision event before comparing, skipping non-decision events
+            while drift_index < len(original_events) and original_events[drift_index].get("event_type") != "decision":
+                drift_index += 1
+            event_dict = {
+                "event_type": "decision",
+                "data": {
+                    "chosen_action": chosen_action,
+                    "action": chosen_action,
+                    "confidence": event.confidence,
+                },
+            }
+            drift = drift_detector.compare(event_dict, drift_index)
+            # Advance to the next decision event in the baseline, skipping non-decision events
+            next_index = drift_index + 1
+            while next_index < len(original_events) and original_events[next_index].get("event_type") != "decision":
+                next_index += 1
+            self._drift_compare_index = next_index
+            if drift is not None:
+                drift_events_list = getattr(self, "_drift_events", None)
+                if drift_events_list is not None:
+                    drift_events_list.append(drift)
+
         return event.id
 
     async def record_tool_call(