working model

src/uipath/_cli/_evals/_evaluate.py

@@ -6,19 +6,22 @@
 
 from uipath._cli._evals._runtime import UiPathEvalContext, UiPathEvalRuntime
 from uipath._events._event_bus import EventBus
+from uipath.tracing import LlmOpsHttpExporter
 
 
 async def evaluate(
     runtime_factory: UiPathRuntimeFactoryProtocol,
     trace_manager: UiPathTraceManager,
     eval_context: UiPathEvalContext,
     event_bus: EventBus,
+    live_tracking_exporter: LlmOpsHttpExporter,
 ) -> UiPathRuntimeResult:
     async with UiPathEvalRuntime(
         factory=runtime_factory,
         context=eval_context,
         trace_manager=trace_manager,
         event_bus=event_bus,
+        live_tracking_exporter=live_tracking_exporter,
     ) as eval_runtime:
         results = await eval_runtime.execute()
         await event_bus.wait_for_all(timeout=10)

src/uipath/_cli/_evals/_progress_reporter.py

@@ -357,6 +357,10 @@ async def update_eval_set_run(
         )
 
     async def handle_create_eval_set_run(self, payload: EvalSetRunCreatedEvent) -> None:
+        logger.info("[TraceID] handle_create_eval_set_run: Event handler triggered")
+        logger.info(
+            f"[TraceID] handle_create_eval_set_run: Received payload with eval_set_run_id={payload.eval_set_run_id}"
+        )
         try:
             self.evaluators = {eval.id: eval for eval in payload.evaluators}
             self.evaluator_scores = {eval.id: [] for eval in payload.evaluators}
@@ -369,6 +373,9 @@ async def handle_create_eval_set_run(self, payload: EvalSetRunCreatedEvent) -> N
             self.is_coded_eval[payload.execution_id] = is_coded
 
             eval_set_run_id = payload.eval_set_run_id
+            logger.info(
+                f"[TraceID] handle_create_eval_set_run: eval_set_run_id from payload = {eval_set_run_id}"
+            )
             if not eval_set_run_id:
                 eval_set_run_id = await self.create_eval_set_run_sw(
                     eval_set_id=payload.eval_set_id,
@@ -422,16 +429,6 @@ async def handle_update_eval_run(self, payload: EvalRunUpdatedEvent) -> None:
         try:
             eval_run_id = self.eval_run_ids.get(payload.execution_id)
 
-            # Use evalRunId as the trace_id for agent execution spans
-            # This makes all agent spans children of the eval run trace
-            if eval_run_id:
-                self.spans_exporter.trace_id = eval_run_id
-            else:
-                # Fallback to evalSetRunId if eval_run_id not available yet
-                if self.eval_set_execution_id:
-                    self.spans_exporter.trace_id = self.eval_set_run_ids.get(
-                        self.eval_set_execution_id
-                    )
 
             self.spans_exporter.export(payload.spans)
 

src/uipath/_cli/_evals/_runtime.py

@@ -308,6 +308,7 @@ def __init__(
         factory: UiPathRuntimeFactoryProtocol,
         trace_manager: UiPathTraceManager,
         event_bus: EventBus,
+        live_tracking_exporter: LlmOpsHttpExporter,
     ):
         self.context: UiPathEvalContext = context
         # Wrap the factory to support model settings overrides
@@ -323,11 +324,15 @@ def __init__(
         self.trace_manager.tracer_provider.add_span_processor(span_processor)
 
         # Live tracking processor for real-time span updates
-        live_tracking_exporter = LlmOpsHttpExporter()
-        live_tracking_processor = LiveTrackingSpanProcessor(live_tracking_exporter)
+        self.live_tracking_exporter = live_tracking_exporter
+        live_tracking_processor = LiveTrackingSpanProcessor(self.live_tracking_exporter)
         self.trace_manager.tracer_span_processors.append(live_tracking_processor)
         self.trace_manager.tracer_provider.add_span_processor(live_tracking_processor)
 
+        logger.info(
+            f"[TraceID] UiPathEvalRuntime.__init__: live_tracking_exporter.trace_id = {getattr(self.live_tracking_exporter, 'trace_id', None)}"
+        )
+
         self.logs_exporter: ExecutionLogsExporter = ExecutionLogsExporter()
         # Use job_id if available (for single runtime runs), otherwise generate UUID
         self.execution_id = context.job_id or str(uuid.uuid4())
@@ -386,6 +391,9 @@ async def initiate_evaluation(
         )
         evaluators = await self._load_evaluators(evaluation_set, runtime)
 
+        logger.info(
+            f"[TraceID] Publishing CREATE_EVAL_SET_RUN event with eval_set_run_id={self.context.eval_set_run_id}"
+        )
         await self.event_bus.publish(
             EvaluationEvents.CREATE_EVAL_SET_RUN,
             EvalSetRunCreatedEvent(
@@ -397,6 +405,9 @@ async def initiate_evaluation(
                 evaluators=evaluators,
             ),
         )
+        logger.info(
+            f"[TraceID] After event publish, exporter.trace_id = {getattr(self.live_tracking_exporter, 'trace_id', None)}"
+        )
 
         return (
             evaluation_set,
@@ -440,9 +451,16 @@ async def execute(self) -> UiPathRuntimeResult:
                 }
                 if self.context.eval_set_run_id:
                     span_attributes["eval_set_run_id"] = self.context.eval_set_run_id
+
+                logger.info(
+                    f"[TraceID] About to create 'Evaluation Set Run' span. Current exporter.trace_id = {getattr(self.live_tracking_exporter, 'trace_id', None)}"
+                )
                 with tracer.start_as_current_span(
                     "Evaluation Set Run", attributes=span_attributes
                 ) as span:
+                    logger.info(
+                        f"[TraceID] Inside 'Evaluation Set Run' span. Exporter.trace_id = {getattr(self.live_tracking_exporter, 'trace_id', None)}"
+                    )
                     try:
                         (
                             evaluation_set,

src/uipath/_cli/cli_eval.py

@@ -1,12 +1,15 @@
 import ast
 import asyncio
+import logging
 import os
 from typing import Any
 
 import click
 from uipath.core.tracing import UiPathTraceManager
 from uipath.runtime import UiPathRuntimeContext, UiPathRuntimeFactoryRegistry
 
+logger = logging.getLogger(__name__)
+
 from uipath._cli._evals._console_progress_reporter import ConsoleProgressReporter
 from uipath._cli._evals._evaluate import evaluate
 from uipath._cli._evals._progress_reporter import StudioWebProgressReporter
@@ -203,8 +206,18 @@ def eval(
             async def execute_eval():
                 event_bus = EventBus()
 
+                live_tracking_exporter = LlmOpsHttpExporter()
+
+                if eval_context.eval_set_run_id:
+                    live_tracking_exporter.trace_id = eval_context.eval_set_run_id
+                    logger.info(
+                        f"[TraceID] Set live_tracking_exporter.trace_id = {eval_context.eval_set_run_id} (user-provided via --eval-set-run-id)"
+                    )
+
                 if should_register_progress_reporter:
-                    progress_reporter = StudioWebProgressReporter(LlmOpsHttpExporter())
+                    progress_reporter = StudioWebProgressReporter(
+                        live_tracking_exporter
+                    )
                     await progress_reporter.subscribe_to_eval_runtime_events(event_bus)
 
                 console_reporter = ConsoleProgressReporter()
@@ -247,11 +260,16 @@ async def execute_eval():
                                     trace_manager,
                                     eval_context,
                                     event_bus,
+                                    live_tracking_exporter,
                                 )
                         else:
                             # Fall back to execution without overwrites
                             ctx.result = await evaluate(
-                                runtime_factory, trace_manager, eval_context, event_bus
+                                runtime_factory,
+                                trace_manager,
+                                eval_context,
+                                event_bus,
+                                live_tracking_exporter,
                             )
                     finally:
                         if runtime_factory: