fix: fix missing details in json output when running uipath eval

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "uipath"
-version = "2.9.13"
+version = "2.9.14"
 description = "Python SDK and CLI for UiPath Platform, enabling programmatic interaction with automation services, process management, and deployment tools."
 readme = { file = "README.md", content-type = "text/markdown" }
 requires-python = ">=3.11"

src/uipath/_cli/cli_eval.py

@@ -205,6 +205,12 @@ def _resolve_model_settings_override(
     default=False,
     help="Resume execution from a previous suspended state",
 )
+@click.option(
+    "--verbose",
+    is_flag=True,
+    default=False,
+    help="Include agent execution output (trace, result) in the output file",
+)
 def eval(
     entrypoint: str | None,
     eval_set: str | None,
@@ -220,6 +226,7 @@ def eval(
     max_llm_concurrency: int,
     input_overrides: dict[str, Any],
     resume: bool,
+    verbose: bool,
 ) -> None:
     """Run an evaluation set against the agent.
 
@@ -272,6 +279,7 @@ def eval(
         eval_context.report_coverage = report_coverage
         eval_context.input_overrides = input_overrides
         eval_context.resume = resume
+        eval_context.verbose = verbose
 
         try:
 

src/uipath/eval/runtime/_types.py

@@ -55,7 +55,7 @@ class EvaluationResultDto(BaseModel):
     model_config = ConfigDict(alias_generator=to_camel, populate_by_name=True)
 
     score: float
-    details: str | BaseModel | None = None
+    details: str | dict[str, Any] | None = None
     evaluation_time: float | None = None
 
     @model_serializer(mode="wrap")
@@ -82,9 +82,17 @@ def from_evaluation_result(
         else:
             score = evaluation_result.score
 
+        # Convert BaseModel details to dict so Pydantic doesn't lose subclass fields
+        if isinstance(evaluation_result.details, BaseModel):
+            details: str | dict[str, Any] | None = (
+                evaluation_result.details.model_dump()
+            )
+        else:
+            details = evaluation_result.details
+
         return cls(
             score=score,
-            details=evaluation_result.details,
+            details=details,
             evaluation_time=evaluation_result.evaluation_time,
         )
 

tests/cli/eval/test_evaluate.py

@@ -120,24 +120,25 @@ async def dispose(self) -> None:
     )
 
     # Assert that the output is json-serializable
-    UiPathEvalOutput.model_validate(result.output).model_dump_json()
+    eval_output = UiPathEvalOutput.model_validate(result.output)
+    eval_output.model_dump_json()
     assert result.output
     output_dict = (
         result.output.model_dump()
         if isinstance(result.output, BaseModel)
         else result.output
     )
     assert isinstance(output_dict, dict)
-    assert (
-        output_dict["evaluationSetResults"][0]["evaluationRunResults"][0]["result"][
-            "score"
-        ]
-        == 1.0
-    )
-    assert (
-        output_dict["evaluationSetResults"][0]["evaluationRunResults"][0]["evaluatorId"]
-        == "ExactMatchEvaluator"
-    )
+    first_result = output_dict["evaluationSetResults"][0]["evaluationRunResults"][0]
+    assert first_result["result"]["score"] == 1.0
+    assert first_result["evaluatorId"] == "ExactMatchEvaluator"
+    # Verify details are properly serialized (not empty dict)
+    details = first_result["result"].get("details")
+    if details is not None:
+        assert details != {}, (
+            "details should not be an empty dict - BaseModel serialization bug"
+        )
+        assert isinstance(details, (str, dict))
 
 
 async def test_eval_runtime_generates_uuid_when_no_custom_id():