diff --git a/distributions/dependencies.json b/distributions/dependencies.json
index d2ed12d3a..33b497a33 100644
--- a/distributions/dependencies.json
+++ b/distributions/dependencies.json
@@ -401,16 +401,13 @@
   ],
   "nvidia": [
     "aiosqlite",
-    "autoevals",
     "blobfile",
     "chardet",
-    "datasets",
     "faiss-cpu",
     "fastapi",
     "fire",
     "httpx",
     "matplotlib",
-    "mcp",
     "nltk",
     "numpy",
     "openai",
diff --git a/docs/source/distributions/remote_hosted_distro/nvidia.md b/docs/source/distributions/remote_hosted_distro/nvidia.md
index efa0a2d74..774d5ec1b 100644
--- a/docs/source/distributions/remote_hosted_distro/nvidia.md
+++ b/docs/source/distributions/remote_hosted_distro/nvidia.md
@@ -6,13 +6,13 @@ The `llamastack/distribution-nvidia` distribution consists of the following prov
 | API | Provider(s) |
 |-----|-------------|
 | agents | `inline::meta-reference` |
-| datasetio | `remote::huggingface`, `inline::localfs` |
+| datasetio | `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::nvidia` |
-| safety | `inline::llama-guard` |
-| scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
+| safety | `remote::nvidia` |
+| scoring | `inline::basic` |
 | telemetry | `inline::meta-reference` |
-| tool_runtime | `remote::brave-search`, `remote::tavily-search`, `inline::code-interpreter`, `inline::rag-runtime`, `remote::model-context-protocol` |
+| tool_runtime | `inline::rag-runtime` |
 | vector_io | `inline::faiss` |
 
 
@@ -20,8 +20,10 @@ The `llamastack/distribution-nvidia` distribution consists of the following prov
 
 The following environment variables can be configured:
 
-- `LLAMASTACK_PORT`: Port for the Llama Stack distribution server (default: `5001`)
 - `NVIDIA_API_KEY`: NVIDIA API Key (default: ``)
+- `GUARDRAILS_SERVICE_URL`: URL for the NeMo Guardrails Service (default: `http://0.0.0.0:7331`)
+- `INFERENCE_MODEL`: Inference model (default: `Llama3.1-8B-Instruct`)
+- `SAFETY_MODEL`: Name of the model to use for safety (default: `meta/llama-3.1-8b-instruct`)
 
 ### Models
 
diff --git a/llama_stack/providers/registry/safety.py b/llama_stack/providers/registry/safety.py
index 1364352e6..32c0b4e98 100644
--- a/llama_stack/providers/registry/safety.py
+++ b/llama_stack/providers/registry/safety.py
@@ -55,4 +55,13 @@ def available_providers() -> List[ProviderSpec]:
                 config_class="llama_stack.providers.remote.safety.bedrock.BedrockSafetyConfig",
             ),
         ),
+        remote_provider_spec(
+            api=Api.safety,
+            adapter=AdapterSpec(
+                adapter_type="nvidia",
+                pip_packages=["requests"],
+                module="llama_stack.providers.remote.safety.nvidia",
+                config_class="llama_stack.providers.remote.safety.nvidia.NVIDIASafetyConfig",
+            ),
+        ),
     ]
diff --git a/llama_stack/providers/remote/safety/nvidia/__init__.py b/llama_stack/providers/remote/safety/nvidia/__init__.py
new file mode 100644
index 000000000..4677268c6
--- /dev/null
+++ b/llama_stack/providers/remote/safety/nvidia/__init__.py
@@ -0,0 +1,18 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+
+from typing import Any
+
+from .config import NVIDIASafetyConfig
+
+
+async def get_adapter_impl(config: NVIDIASafetyConfig, _deps) -> Any:
+    from .nvidia import NVIDIASafetyAdapter
+
+    impl = NVIDIASafetyAdapter(config)
+    await impl.initialize()
+    return impl
diff --git a/llama_stack/providers/remote/safety/nvidia/config.py b/llama_stack/providers/remote/safety/nvidia/config.py
new file mode 100644
index 000000000..3df80ed4f
--- /dev/null
+++ b/llama_stack/providers/remote/safety/nvidia/config.py
@@ -0,0 +1,37 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+import os
+from typing import Any, Dict, Optional
+
+from pydantic import BaseModel, Field
+
+from llama_stack.schema_utils import json_schema_type
+
+
+@json_schema_type
+class NVIDIASafetyConfig(BaseModel):
+    """
+    Configuration for the NVIDIA Guardrail microservice endpoint.
+
+    Attributes:
+        guardrails_service_url (str): A base url for accessing the NVIDIA guardrail endpoint, e.g. http://0.0.0.0:7331
+        config_id (str): The ID of the guardrails configuration to use from the configuration store
+         (https://developer.nvidia.com/docs/nemo-microservices/guardrails/source/guides/configuration-store-guide.html)
+
+    """
+
+    guardrails_service_url: str = Field(
+        default_factory=lambda: os.getenv("GUARDRAILS_SERVICE_URL", "http://0.0.0.0:7331"),
+        description="The url for accessing the guardrails service",
+    )
+    config_id: Optional[str] = Field(default="self-check", description="Config ID to use from the config store")
+
+    @classmethod
+    def sample_run_config(cls, **kwargs) -> Dict[str, Any]:
+        return {
+            "guardrails_service_url": "${env.GUARDRAILS_SERVICE_URL:http://localhost:7331}",
+            "config_id": "self-check",
+        }
diff --git a/llama_stack/providers/remote/safety/nvidia/nvidia.py b/llama_stack/providers/remote/safety/nvidia/nvidia.py
new file mode 100644
index 000000000..6da2a8344
--- /dev/null
+++ b/llama_stack/providers/remote/safety/nvidia/nvidia.py
@@ -0,0 +1,154 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+import logging
+from typing import Any, List, Optional
+
+import requests
+
+from llama_stack.apis.inference import Message
+from llama_stack.apis.safety import RunShieldResponse, Safety, SafetyViolation, ViolationLevel
+from llama_stack.apis.shields import Shield
+from llama_stack.distribution.library_client import convert_pydantic_to_json_value
+from llama_stack.providers.datatypes import ShieldsProtocolPrivate
+
+from .config import NVIDIASafetyConfig
+
+logger = logging.getLogger(__name__)
+
+
+class NVIDIASafetyAdapter(Safety, ShieldsProtocolPrivate):
+    def __init__(self, config: NVIDIASafetyConfig) -> None:
+        """
+        Initialize the NVIDIASafetyAdapter with a given safety configuration.
+
+        Args:
+            config (NVIDIASafetyConfig): The configuration containing the guardrails service URL and config ID.
+        """
+        print(f"Initializing NVIDIASafetyAdapter({config.guardrails_service_url})...")
+        self.config = config
+
+    async def initialize(self) -> None:
+        pass
+
+    async def shutdown(self) -> None:
+        pass
+
+    async def register_shield(self, shield: Shield) -> None:
+        if not shield.provider_resource_id:
+            raise ValueError("Shield model not provided.")
+
+    async def run_shield(
+        self, shield_id: str, messages: List[Message], params: Optional[dict[str, Any]] = None
+    ) -> RunShieldResponse:
+        """
+        Run a safety shield check against the provided messages.
+
+        Args:
+            shield_id (str): The unique identifier for the shield to be used.
+            messages (List[Message]): A list of Message objects representing the conversation history.
+            params (Optional[dict[str, Any]]): Additional parameters for the shield check.
+
+        Returns:
+            RunShieldResponse: The response containing safety violation details if any.
+
+        Raises:
+            ValueError: If the shield with the provided shield_id is not found.
+        """
+        shield = await self.shield_store.get_shield(shield_id)
+        if not shield:
+            raise ValueError(f"Shield {shield_id} not found")
+
+        self.shield = NeMoGuardrails(self.config, shield.shield_id)
+        return await self.shield.run(messages)
+
+
+class NeMoGuardrails:
+    """
+    A class that encapsulates NVIDIA's guardrails safety logic.
+
+    Sends messages to the guardrails service and interprets the response to determine
+    if a safety violation has occurred.
+    """
+
+    def __init__(
+        self,
+        config: NVIDIASafetyConfig,
+        model: str,
+        threshold: float = 0.9,
+        temperature: float = 1.0,
+    ):
+        """
+        Initialize a NeMoGuardrails instance with the provided parameters.
+
+        Args:
+            config (NVIDIASafetyConfig): The safety configuration containing the config ID and guardrails URL.
+            model (str): The identifier or name of the model to be used for safety checks.
+            threshold (float, optional): The threshold for flagging violations. Defaults to 0.9.
+            temperature (float, optional): The temperature setting for the underlying model. Must be greater than 0. Defaults to 1.0.
+
+        Raises:
+            ValueError: If temperature is less than or equal to 0.
+            AssertionError: If config_id is not provided in the configuration.
+        """
+        self.config_id = config.config_id
+        self.model = model
+        assert self.config_id is not None, "Must provide config id"
+        if temperature <= 0:
+            raise ValueError("Temperature must be greater than 0")
+
+        self.temperature = temperature
+        self.threshold = threshold
+        self.guardrails_service_url = config.guardrails_service_url
+
+    async def run(self, messages: List[Message]) -> RunShieldResponse:
+        """
+        Queries the /v1/guardrails/checks endpoint of the NeMo guardrails deployed API.
+
+        Args:
+            messages (List[Message]): A list of Message objects to be checked for safety violations.
+
+        Returns:
+            RunShieldResponse: If the response indicates a violation ("blocked" status), returns a
+            RunShieldResponse with a SafetyViolation; otherwise, returns a RunShieldResponse with violation set to None.
+
+        Raises:
+            requests.HTTPError: If the POST request fails.
+        """
+        headers = {
+            "Accept": "application/json",
+        }
+        request_data = {
+            "model": self.model,
+            "messages": convert_pydantic_to_json_value(messages),
+            "temperature": self.temperature,
+            "top_p": 1,
+            "frequency_penalty": 0,
+            "presence_penalty": 0,
+            "max_tokens": 160,
+            "stream": False,
+            "guardrails": {
+                "config_id": self.config_id,
+            },
+        }
+        response = requests.post(
+            url=f"{self.guardrails_service_url}/v1/guardrail/checks", headers=headers, json=request_data
+        )
+        response.raise_for_status()
+        if "Content-Type" in response.headers and response.headers["Content-Type"].startswith("application/json"):
+            response_json = response.json()
+        if response_json["status"] == "blocked":
+            user_message = "Sorry I cannot do this."
+            metadata = response_json["rails_status"]
+
+            return RunShieldResponse(
+                violation=SafetyViolation(
+                    user_message=user_message,
+                    violation_level=ViolationLevel.ERROR,
+                    metadata=metadata,
+                )
+            )
+        return RunShieldResponse(violation=None)
diff --git a/llama_stack/templates/nvidia/build.yaml b/llama_stack/templates/nvidia/build.yaml
index e9748721a..0c788ce86 100644
--- a/llama_stack/templates/nvidia/build.yaml
+++ b/llama_stack/templates/nvidia/build.yaml
@@ -1,13 +1,13 @@
 version: '2'
 distribution_spec:
-  description: Use NVIDIA NIM for running LLM inference
+  description: Use NVIDIA NIM for running LLM inference and safety
   providers:
     inference:
     - remote::nvidia
     vector_io:
     - inline::faiss
     safety:
-    - inline::llama-guard
+    - remote::nvidia
     agents:
     - inline::meta-reference
     telemetry:
@@ -15,16 +15,9 @@ distribution_spec:
     eval:
     - inline::meta-reference
     datasetio:
-    - remote::huggingface
     - inline::localfs
     scoring:
     - inline::basic
-    - inline::llm-as-judge
-    - inline::braintrust
     tool_runtime:
-    - remote::brave-search
-    - remote::tavily-search
-    - inline::code-interpreter
     - inline::rag-runtime
-    - remote::model-context-protocol
 image_type: conda
diff --git a/llama_stack/templates/nvidia/nvidia.py b/llama_stack/templates/nvidia/nvidia.py
index cc5e96333..308c0e2a6 100644
--- a/llama_stack/templates/nvidia/nvidia.py
+++ b/llama_stack/templates/nvidia/nvidia.py
@@ -6,9 +6,10 @@
 
 from pathlib import Path
 
-from llama_stack.distribution.datatypes import Provider, ToolGroupInput
+from llama_stack.distribution.datatypes import ModelInput, Provider, ShieldInput, ToolGroupInput
 from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig
 from llama_stack.providers.remote.inference.nvidia.models import MODEL_ENTRIES
+from llama_stack.providers.remote.safety.nvidia import NVIDIASafetyConfig
 from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
 
 
@@ -16,19 +17,13 @@ def get_distribution_template() -> DistributionTemplate:
     providers = {
         "inference": ["remote::nvidia"],
         "vector_io": ["inline::faiss"],
-        "safety": ["inline::llama-guard"],
+        "safety": ["remote::nvidia"],
         "agents": ["inline::meta-reference"],
         "telemetry": ["inline::meta-reference"],
         "eval": ["inline::meta-reference"],
-        "datasetio": ["remote::huggingface", "inline::localfs"],
-        "scoring": ["inline::basic", "inline::llm-as-judge", "inline::braintrust"],
-        "tool_runtime": [
-            "remote::brave-search",
-            "remote::tavily-search",
-            "inline::code-interpreter",
-            "inline::rag-runtime",
-            "remote::model-context-protocol",
-        ],
+        "datasetio": ["inline::localfs"],
+        "scoring": ["inline::basic"],
+        "tool_runtime": ["inline::rag-runtime"],
     }
 
     inference_provider = Provider(
@@ -36,30 +31,35 @@ def get_distribution_template() -> DistributionTemplate:
         provider_type="remote::nvidia",
         config=NVIDIAConfig.sample_run_config(),
     )
+    safety_provider = Provider(
+        provider_id="nvidia",
+        provider_type="remote::nvidia",
+        config=NVIDIASafetyConfig.sample_run_config(),
+    )
+    inference_model = ModelInput(
+        model_id="${env.INFERENCE_MODEL}",
+        provider_id="nvidia",
+    )
+    safety_model = ModelInput(
+        model_id="${env.SAFETY_MODEL}",
+        provider_id="nvidia",
+    )
 
     available_models = {
         "nvidia": MODEL_ENTRIES,
     }
     default_tool_groups = [
-        ToolGroupInput(
-            toolgroup_id="builtin::websearch",
-            provider_id="tavily-search",
-        ),
         ToolGroupInput(
             toolgroup_id="builtin::rag",
             provider_id="rag-runtime",
         ),
-        ToolGroupInput(
-            toolgroup_id="builtin::code_interpreter",
-            provider_id="code-interpreter",
-        ),
     ]
 
     default_models = get_model_registry(available_models)
     return DistributionTemplate(
         name="nvidia",
         distro_type="remote_hosted",
-        description="Use NVIDIA NIM for running LLM inference",
+        description="Use NVIDIA NIM for running LLM inference and safety",
         container_image=None,
         template_path=Path(__file__).parent / "doc_template.md",
         providers=providers,
@@ -72,15 +72,34 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=default_models,
                 default_tool_groups=default_tool_groups,
             ),
+            "run-with-safety.yaml": RunConfigSettings(
+                provider_overrides={
+                    "inference": [
+                        inference_provider,
+                        safety_provider,
+                    ]
+                },
+                default_models=[inference_model, safety_model],
+                default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}", provider_id="nvidia")],
+                default_tool_groups=default_tool_groups,
+            ),
         },
         run_config_env_vars={
-            "LLAMASTACK_PORT": (
-                "5001",
-                "Port for the Llama Stack distribution server",
-            ),
             "NVIDIA_API_KEY": (
                 "",
                 "NVIDIA API Key",
             ),
+            "GUARDRAILS_SERVICE_URL": (
+                "http://0.0.0.0:7331",
+                "URL for the NeMo Guardrails Service",
+            ),
+            "INFERENCE_MODEL": (
+                "Llama3.1-8B-Instruct",
+                "Inference model",
+            ),
+            "SAFETY_MODEL": (
+                "meta/llama-3.1-8b-instruct",
+                "Name of the model to use for safety",
+            ),
         },
     )
diff --git a/llama_stack/templates/nvidia/run-with-safety.yaml b/llama_stack/templates/nvidia/run-with-safety.yaml
new file mode 100644
index 000000000..04da1bcda
--- /dev/null
+++ b/llama_stack/templates/nvidia/run-with-safety.yaml
@@ -0,0 +1,101 @@
+version: '2'
+image_name: nvidia
+apis:
+- agents
+- datasetio
+- eval
+- inference
+- safety
+- scoring
+- telemetry
+- tool_runtime
+- vector_io
+providers:
+  inference:
+  - provider_id: nvidia
+    provider_type: remote::nvidia
+    config:
+      url: ${env.NVIDIA_BASE_URL:https://integrate.api.nvidia.com}
+      api_key: ${env.NVIDIA_API_KEY:}
+  - provider_id: nvidia
+    provider_type: remote::nvidia
+    config:
+      guardrails_service_url: ${env.GUARDRAILS_SERVICE_URL:http://localhost:7331}
+      config_id: self-check
+  vector_io:
+  - provider_id: faiss
+    provider_type: inline::faiss
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/faiss_store.db
+  safety:
+  - provider_id: nvidia
+    provider_type: remote::nvidia
+    config:
+      guardrails_service_url: ${env.GUARDRAILS_SERVICE_URL:http://localhost:7331}
+      config_id: self-check
+  agents:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      persistence_store:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/agents_store.db
+  telemetry:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
+      sinks: ${env.TELEMETRY_SINKS:console,sqlite}
+      sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/nvidia/trace_store.db}
+  eval:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/meta_reference_eval.db
+  datasetio:
+  - provider_id: localfs
+    provider_type: inline::localfs
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/localfs_datasetio.db
+  scoring:
+  - provider_id: basic
+    provider_type: inline::basic
+    config: {}
+  tool_runtime:
+  - provider_id: rag-runtime
+    provider_type: inline::rag-runtime
+    config: {}
+metadata_store:
+  type: sqlite
+  db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
+models:
+- metadata: {}
+  model_id: ${env.INFERENCE_MODEL}
+  provider_id: nvidia
+  model_type: llm
+- metadata: {}
+  model_id: ${env.SAFETY_MODEL}
+  provider_id: nvidia
+  model_type: llm
+shields:
+- shield_id: ${env.SAFETY_MODEL}
+  provider_id: nvidia
+vector_dbs: []
+datasets: []
+scoring_fns: []
+benchmarks: []
+tool_groups:
+- toolgroup_id: builtin::rag
+  provider_id: rag-runtime
+server:
+  port: 8321
diff --git a/llama_stack/templates/nvidia/run.yaml b/llama_stack/templates/nvidia/run.yaml
index 213e22cb2..3abdd82a7 100644
--- a/llama_stack/templates/nvidia/run.yaml
+++ b/llama_stack/templates/nvidia/run.yaml
@@ -26,10 +26,11 @@ providers:
         namespace: null
         db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/faiss_store.db
   safety:
-  - provider_id: llama-guard
-    provider_type: inline::llama-guard
+  - provider_id: nvidia
+    provider_type: remote::nvidia
     config:
-      excluded_categories: []
+      guardrails_service_url: ${env.GUARDRAILS_SERVICE_URL:http://localhost:7331}
+      config_id: self-check
   agents:
   - provider_id: meta-reference
     provider_type: inline::meta-reference
@@ -54,13 +55,6 @@ providers:
         namespace: null
         db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/meta_reference_eval.db
   datasetio:
-  - provider_id: huggingface
-    provider_type: remote::huggingface
-    config:
-      kvstore:
-        type: sqlite
-        namespace: null
-        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/huggingface_datasetio.db
   - provider_id: localfs
     provider_type: inline::localfs
     config:
@@ -72,33 +66,10 @@ providers:
   - provider_id: basic
     provider_type: inline::basic
     config: {}
-  - provider_id: llm-as-judge
-    provider_type: inline::llm-as-judge
-    config: {}
-  - provider_id: braintrust
-    provider_type: inline::braintrust
-    config:
-      openai_api_key: ${env.OPENAI_API_KEY:}
   tool_runtime:
-  - provider_id: brave-search
-    provider_type: remote::brave-search
-    config:
-      api_key: ${env.BRAVE_SEARCH_API_KEY:}
-      max_results: 3
-  - provider_id: tavily-search
-    provider_type: remote::tavily-search
-    config:
-      api_key: ${env.TAVILY_SEARCH_API_KEY:}
-      max_results: 3
-  - provider_id: code-interpreter
-    provider_type: inline::code-interpreter
-    config: {}
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  - provider_id: model-context-protocol
-    provider_type: remote::model-context-protocol
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
@@ -227,11 +198,7 @@ datasets: []
 scoring_fns: []
 benchmarks: []
 tool_groups:
-- toolgroup_id: builtin::websearch
-  provider_id: tavily-search
 - toolgroup_id: builtin::rag
   provider_id: rag-runtime
-- toolgroup_id: builtin::code_interpreter
-  provider_id: code-interpreter
 server:
   port: 8321
diff --git a/pyproject.toml b/pyproject.toml
index 4a5befbd0..a006d69f9 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -269,6 +269,7 @@ exclude = [
     "^llama_stack/providers/remote/inference/together/",
     "^llama_stack/providers/remote/inference/vllm/",
     "^llama_stack/providers/remote/safety/bedrock/",
+    "^llama_stack/providers/remote/safety/nvidia/",
     "^llama_stack/providers/remote/safety/sample/",
     "^llama_stack/providers/remote/tool_runtime/bing_search/",
     "^llama_stack/providers/remote/tool_runtime/brave_search/",