fix(evals): Change eval set discrimination strategy

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "uipath"
-version = "2.6.22"
+version = "2.6.23"
 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"
@@ -22,6 +22,7 @@ dependencies = [
   "coverage>=7.8.2",
   "mermaid-builder==0.0.3",
   "applicationinsights>=0.11.10",
+  "pydantic>=2.12.5",
 ]
 classifiers = [
   "Intended Audience :: Developers",

src/uipath/_cli/_evals/_models/_evaluation_set.py

@@ -186,14 +186,3 @@ class EvaluationStatus(IntEnum):
     IN_PROGRESS = 1
     COMPLETED = 2
     FAILED = 3
-
-
-def _discriminate_eval_set(
-    v: Any,
-) -> Literal["evaluation_set", "legacy_evaluation_set"]:
-    """Discriminator function that returns a tag based on version field."""
-    if isinstance(v, dict):
-        version = v.get("version")
-        if version == "1.0":
-            return "evaluation_set"
-    return "legacy_evaluation_set"

src/uipath/_cli/_utils/_eval_set.py

@@ -1,8 +1,9 @@
 import json
 from pathlib import Path
+from typing import Any
 
 import click
-from pydantic import TypeAdapter, ValidationError
+from pydantic import ValidationError
 
 from uipath._cli._evals._models._evaluation_set import (
     EvaluationItem,
@@ -19,6 +20,24 @@
 EVAL_SETS_DIRECTORY_NAME = "evaluations/eval-sets"
 
 
+def discriminate_eval_set(data: dict[str, Any]) -> EvaluationSet | LegacyEvaluationSet:
+    """Discriminate and parse evaluation set based on version field.
+
+    Uses explicit version checking instead of Pydantic's smart union matching
+    to avoid incorrect type selection when both types have matching fields.
+
+    Args:
+        data: Dictionary containing evaluation set data
+
+    Returns:
+        Either EvaluationSet (for version 1.0) or LegacyEvaluationSet
+    """
+    version = data.get("version")
+    if version == "1.0":
+        return EvaluationSet.model_validate(data)
+    return LegacyEvaluationSet.model_validate(data)
+
+
 class EvalHelpers:
     @staticmethod
     def auto_discover_eval_set() -> str:
@@ -100,9 +119,7 @@ def load_eval_set(
             ) from e
 
         try:
-            eval_set: EvaluationSet | LegacyEvaluationSet = TypeAdapter(
-                EvaluationSet | LegacyEvaluationSet
-            ).validate_python(data)
+            eval_set = discriminate_eval_set(data)
             if isinstance(eval_set, LegacyEvaluationSet):
 
                 def migrate_evaluation_item(