diff --git a/litellm/proxy/guardrails/guardrail_hooks/promptguard/promptguard.py b/litellm/proxy/guardrails/guardrail_hooks/promptguard/promptguard.py
index 009e29ef089..8c5001249ad 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/promptguard/promptguard.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/promptguard/promptguard.py
@@ -155,7 +155,13 @@ class PromptGuardGuardrail(CustomGuardrail):
         except Exception as exc:
             verbose_proxy_logger.error("PromptGuard API error: %s", str(exc))
             if self.block_on_error:
-                raise
+                raise GuardrailRaisedException(
+                    guardrail_name=self.guardrail_name,
+                    message=(
+                        f"PromptGuard API unreachable "
+                        f"(block_on_error=True): {exc}"
+                    ),
+                ) from exc
             return inputs
 
         verbose_proxy_logger.debug(
@@ -164,7 +170,7 @@ class PromptGuardGuardrail(CustomGuardrail):
             result.get("threat_type"),
         )
 
-        decision = result.get("decision", "allow")
+        decision = result.get("decision") or "allow"
 
         if decision == "block":
             threat_type = result.get("threat_type", "unknown")
@@ -196,9 +202,16 @@ class PromptGuardGuardrail(CustomGuardrail):
 
     @staticmethod
     def _extract_texts_from_messages(messages: list) -> List[str]:
-        """Extract text content strings from a list of chat messages."""
+        """Extract text content from user-role messages only.
+
+        Only user messages are extracted to avoid injecting system or
+        assistant content into the ``texts`` list, which should mirror
+        the original user-provided input.
+        """
         texts: List[str] = []
         for message in messages:
+            if message.get("role") != "user":
+                continue
             content = message.get("content")
             if isinstance(content, str):
                 texts.append(content)
diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_promptguard.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_promptguard.py
index e5e2face752..efd14379ddd 100644
--- a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_promptguard.py
+++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_promptguard.py
@@ -373,7 +373,7 @@ class TestPromptGuardRedactAction:
                 input_type="request",
             )
             assert result["structured_messages"] == redacted
-            assert "My SSN is *********" in result["texts"]
+            assert result["texts"] == ["My SSN is *********"]
 
     @pytest.mark.asyncio
     async def test_redact_structured_only_does_not_create_texts(
@@ -629,7 +629,7 @@ class TestPromptGuardErrorHandling:
     async def test_http_error_propagates_block_on_error(
         self, promptguard_guardrail, mock_request_data
     ):
-        """Default block_on_error=True re-raises HTTP errors."""
+        """Default block_on_error=True wraps HTTP errors in GuardrailRaisedException."""
         mock_request = httpx.Request("POST", "https://api.test.promptguard.co")
         mock_resp = httpx.Response(status_code=500, request=mock_request)
         with patch.object(
@@ -641,29 +641,33 @@ class TestPromptGuardErrorHandling:
                 response=mock_resp,
             ),
         ):
-            with pytest.raises(httpx.HTTPStatusError):
+            with pytest.raises(GuardrailRaisedException) as exc_info:
                 await promptguard_guardrail.apply_guardrail(
                     inputs={"texts": ["test"]},
                     request_data=mock_request_data,
                     input_type="request",
                 )
+            assert "block_on_error=True" in str(exc_info.value)
+            assert exc_info.value.__cause__ is not None
 
     @pytest.mark.asyncio
     async def test_connection_error_propagates_block_on_error(
         self, promptguard_guardrail, mock_request_data
     ):
-        """Default block_on_error=True re-raises connection errors."""
+        """Default block_on_error=True wraps connection errors in GuardrailRaisedException."""
         with patch.object(
             promptguard_guardrail.async_handler,
             "post",
             side_effect=httpx.ConnectError("Connection refused"),
         ):
-            with pytest.raises(httpx.ConnectError):
+            with pytest.raises(GuardrailRaisedException) as exc_info:
                 await promptguard_guardrail.apply_guardrail(
                     inputs={"texts": ["test"]},
                     request_data=mock_request_data,
                     input_type="request",
                 )
+            assert "block_on_error=True" in str(exc_info.value)
+            assert exc_info.value.__cause__ is not None
 
     @pytest.mark.asyncio
     async def test_fail_open_returns_inputs_on_http_error(self, mock_request_data):
@@ -747,6 +751,22 @@ class TestPromptGuardErrorHandling:
             )
             assert result["texts"] == ["test"]
 
+    @pytest.mark.asyncio
+    async def test_null_decision_treated_as_allow(
+        self, promptguard_guardrail, mock_request_data
+    ):
+        """Explicit null decision should be treated as allow."""
+        resp = _make_response({"decision": None, "event_id": "evt-null"})
+        with patch.object(
+            promptguard_guardrail.async_handler, "post", return_value=resp
+        ):
+            result = await promptguard_guardrail.apply_guardrail(
+                inputs={"texts": ["test"]},
+                request_data=mock_request_data,
+                input_type="request",
+            )
+            assert result["texts"] == ["test"]
+
 
 # ---------------------------------------------------------------------------
 # Config model
diff --git a/ui/litellm-dashboard/src/components/guardrails/guardrail_info_helpers.tsx b/ui/litellm-dashboard/src/components/guardrails/guardrail_info_helpers.tsx
index 088995dea8d..8ab8710d01a 100644
--- a/ui/litellm-dashboard/src/components/guardrails/guardrail_info_helpers.tsx
+++ b/ui/litellm-dashboard/src/components/guardrails/guardrail_info_helpers.tsx
@@ -48,6 +48,7 @@ export const guardrail_provider_map: Record<string, string> = {
   LitellmContentFilter: "litellm_content_filter",
   ToolPermission: "tool_permission",
   BlockCodeExecution: "block_code_execution",
+  Promptguard: "promptguard",
 };
 
 // Function to populate provider map from API response - updates the original map