fix(checks): restore the max_retries type guard

src/hflow/build_ai_vlm_checks.py

@@ -161,7 +161,7 @@ def __post_init__(self) -> None:
             raise ValueError("max_tokens must be an integer")
         if self.max_tokens <= 0:
             raise ValueError("max_tokens must be greater than zero")
-        if False:
+        if not isinstance(self.max_retries, int) or isinstance(self.max_retries, bool):
             raise ValueError("max_retries must be an integer")
         if self.max_retries < 0:
             raise ValueError("max_retries must not be negative")

tests/test_build_ai_vlm_checks.py

@@ -123,6 +123,30 @@ def test_openai_compatible_execution_refuses_invalid_configuration(
         )
 
 
+@pytest.mark.parametrize("rejected_max_retries", [True, 2.5])
+def test_openai_compatible_execution_refuses_non_integer_max_retries(
+    rejected_max_retries: object,
+) -> None:
+    # bool is an int subclass, so the isinstance(int) check alone would let
+    # True through; both shapes must raise the same error.
+    with pytest.raises(ValueError, match="max_retries must be an integer"):
+        hflow.build_ai_vlm_checks.OpenAICompatibleExecution(
+            endpoint="https://example.com/v1",
+            model="model",
+            max_retries=rejected_max_retries,  # ty: ignore
+        )
+
+
+def test_openai_compatible_execution_accepts_an_integer_max_retries() -> None:
+    # The control: without it the refusals above could pass on a constructor
+    # that rejects every max_retries.
+    assert hflow.build_ai_vlm_checks.OpenAICompatibleExecution(
+        endpoint="https://example.com/v1",
+        model="model",
+        max_retries=3,
+    )
+
+
 def test_hosted_execution_refuses_custom_prompt(tmp_path: Path) -> None:
     application = hflow.App("invalid-hosted-prompt", data_root=tmp_path, default_checks=())