diff --git a/docs/openapi.json b/docs/openapi.json
index ed9b0b96..ecd56e4d 100644
--- a/docs/openapi.json
+++ b/docs/openapi.json
@@ -819,6 +819,13 @@
             "example": "1.0",
             "nullable": true
           },
+          "guideline": {
+            "type": "string",
+            "description": "A guideline to be used in the chat_template",
+            "default": "null",
+            "example": "null",
+            "nullable": true
+          },
           "logit_bias": {
             "type": "array",
             "items": {
diff --git a/router/src/infer/chat_template.rs b/router/src/infer/chat_template.rs
index 9203c1b2..7c2753ed 100644
--- a/router/src/infer/chat_template.rs
+++ b/router/src/infer/chat_template.rs
@@ -733,6 +733,19 @@ mod tests {
                 },
                 target: "You are a friendly chatbot who always responds in the style of a pirateYou are a friendly chatbot who always responds in the style of a pirate### Instruction: Hello, how are you?### Response: I'm doing great. How can I help you today?### Instruction: I'd like to show off how chat templating works!",
             },
+            ChatTemplateTestItem {
+                name: "google/shieldgemma-9b",
+                chat_template: "{{- bos_token }}\n{%- if messages[-1].role == 'user' %}\n    {{- \"You are a policy expert trying to help determine whether a user prompt is in violation of the defined safety policies.\\n\\n\" }}\n    {{- \"<start_of_turn>\\n\" }}\n    {{- \"Human Question: \" + messages[-1].content }}\n    {{- \"\\n<end_of_turn>\\n\\n\" }}\n    {{- \"Our safety principle is defined in the below:\\n\\n\" }}\n    {{- \"* \" + guideline + \"\\n\" }}\n    {{- \"\\n===\\n\\n\" }}\n    {{- \"Does the human question violate the above principle? Your answer must start with 'Yes' or 'No'. And then walk through step by step to be sure we answer correctly.\\n\\n\" }}\n{%- else %}\n    {{- \"You are a policy expert trying to help determine whether an AI response to a prompt is in violation of the defined safety policies.\\n\\n\" }}\n    {{- \"<start_of_turn>\\n\" }}\n    {{- \"Human Question: \" + messages[-2].content }}\n    {{- \"\\n<end_of_turn>\\n\" }}\n    {{- \"<start_of_turn>\\n\" }}\n    {{- \"Chatbot Response: \" + messages[-1].content }}\n    {{- \"\\n<end_of_turn>\\n\\n\" }}\n    {{- \"Our safety principle is defined in the below:\\n\\n\" }}\n    {{- \"* \" + guideline + \"\\n\" }}\n    {{- \"\\n===\\n\\n\" }}\n    {{- \"Does the Chatbot Response violate the above principle? Your answer must  start with 'Yes' or 'No'. And then walk through step by step to be sure we answer correctly.\\n\\n\" }}\n{%- endif %}\n\n",
+                input: ChatTemplateInputs {
+                    messages: example_chat_with_system.clone(),
+                    add_generation_prompt: false,
+                    bos_token: Some("<s>"),
+                    eos_token: Some("</s>"),
+                    guideline: Some("Do not use offensive language."),
+                    ..Default::default()
+                },
+                target: "<s>You are a policy expert trying to help determine whether a user prompt is in violation of the defined safety policies.\n\n<start_of_turn>\nHuman Question: I'd like to show off how chat templating works!\n<end_of_turn>\n\nOur safety principle is defined in the below:\n\n* Do not use offensive language.\n\n===\n\nDoes the human question violate the above principle? Your answer must start with 'Yes' or 'No'. And then walk through step by step to be sure we answer correctly.\n\n",
+            },
         ];
 
         #[allow(unused_variables)] // name is unused