Update provider_type -> inline::llama-guard in templates, update run.yaml

llama_stack/providers/tests/safety/fixtures.py

@@ -11,6 +11,7 @@ from llama_stack.apis.shields import ShieldType
 
 from llama_stack.distribution.datatypes import Api, Provider
 from llama_stack.providers.inline.safety.llama_guard import LlamaGuardConfig
+from llama_stack.providers.inline.safety.prompt_guard import PromptGuardConfig
 from llama_stack.providers.remote.safety.bedrock import BedrockSafetyConfig
 
 from llama_stack.providers.tests.resolver import resolve_impls_for_test_v2
@@ -44,6 +45,22 @@ def safety_llama_guard(safety_model) -> ProviderFixture:
     )
 
 
+# TODO: this is not tested yet; we would need to configure the run_shield() test
+# and parametrize it with the "prompt" for testing depending on the safety fixture
+# we are using.
+@pytest.fixture(scope="session")
+def safety_prompt_guard() -> ProviderFixture:
+    return ProviderFixture(
+        providers=[
+            Provider(
+                provider_id="inline::prompt-guard",
+                provider_type="inline::prompt-guard",
+                config=PromptGuardConfig().model_dump(),
+            )
+        ],
+    )
+
+
 @pytest.fixture(scope="session")
 def safety_bedrock() -> ProviderFixture:
     return ProviderFixture(

llama_stack/templates/bedrock/build.yaml

@@ -3,7 +3,7 @@ distribution_spec:
   description: Use Amazon Bedrock APIs.
   providers:
     inference: remote::bedrock
-    memory: meta-reference
-    safety: meta-reference
+    memory: inline::faiss
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/databricks/build.yaml

@@ -3,7 +3,7 @@ distribution_spec:
   description: Use Databricks for running LLM inference
   providers:
     inference: remote::databricks
-    memory: meta-reference
-    safety: meta-reference
+    memory: inline::faiss
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/fireworks/build.yaml

@@ -6,6 +6,6 @@ distribution_spec:
     memory:
     - meta-reference
     - remote::weaviate
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/hf-endpoint/build.yaml

@@ -3,7 +3,7 @@ distribution_spec:
   description: "Like local, but use Hugging Face Inference Endpoints for running LLM inference.\nSee https://hf.co/docs/api-endpoints."
   providers:
     inference: remote::hf::endpoint
-    memory: meta-reference
-    safety: meta-reference
+    memory: inline::faiss
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/hf-serverless/build.yaml

@@ -3,7 +3,7 @@ distribution_spec:
   description: "Like local, but use Hugging Face Inference API (serverless) for running LLM inference.\nSee https://hf.co/docs/api-inference."
   providers:
     inference: remote::hf::serverless
-    memory: meta-reference
-    safety: meta-reference
+    memory: inline::faiss
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/inline-vllm/build.yaml

@@ -8,6 +8,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/meta-reference-gpu/build.yaml

@@ -8,6 +8,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/meta-reference-quantized-gpu/build.yaml

@@ -8,6 +8,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/ollama/build.yaml

@@ -7,6 +7,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/remote-vllm/build.yaml

@@ -7,6 +7,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/tgi/build.yaml

@@ -7,6 +7,6 @@ distribution_spec:
     - meta-reference
     - remote::chromadb
     - remote::pgvector
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference

llama_stack/templates/together/build.yaml

@@ -6,6 +6,6 @@ distribution_spec:
     memory:
     - meta-reference
     - remote::weaviate
-    safety: meta-reference
+    safety: inline::llama-guard
     agents: meta-reference
     telemetry: meta-reference