feat: split API and provider specs into separate llama-stack-api pkg (#3895)

# What does this PR do?

Extract API definitions and provider specifications into a standalone
llama-stack-api package that can be published to PyPI independently of
the main llama-stack server.


see: https://github.com/llamastack/llama-stack/pull/2978 and
https://github.com/llamastack/llama-stack/pull/2978#issuecomment-3145115942

Motivation

External providers currently import from llama-stack, which overrides
the installed version and causes dependency conflicts. This separation
allows external providers to:

- Install only the type definitions they need without server
dependencies
- Avoid version conflicts with the installed llama-stack package
- Be versioned and released independently

This enables us to re-enable external provider module tests that were
previously blocked by these import conflicts.

Changes

- Created llama-stack-api package with minimal dependencies (pydantic,
jsonschema)
- Moved APIs, providers datatypes, strong_typing, and schema_utils
- Updated all imports from llama_stack.* to llama_stack_api.*
- Configured local editable install for development workflow
- Updated linting and type-checking configuration for both packages

Next Steps

- Publish llama-stack-api to PyPI
- Update external provider dependencies
- Re-enable external provider module tests


Pre-cursor PRs to this one:

- #4093 
- #3954 
- #4064 

These PRs moved key pieces _out_ of the Api pkg, limiting the scope of
change here.


relates to #3237 

## Test Plan

Package builds successfully and can be imported independently. All
pre-commit hooks pass with expected exclusions maintained.

---------

Signed-off-by: Charlie Doern <cdoern@redhat.com>

src/llama_stack/providers/remote/datasetio/huggingface/huggingface.py

@@ -6,10 +6,8 @@
 from typing import Any
 from urllib.parse import parse_qs, urlparse
 
-from llama_stack.apis.common.responses import PaginatedResponse
-from llama_stack.apis.datasetio import DatasetIO
-from llama_stack.apis.datasets import Dataset
-from llama_stack.providers.datatypes import DatasetsProtocolPrivate
+from llama_stack_api import Dataset, DatasetIO, DatasetsProtocolPrivate, PaginatedResponse
+
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.pagination import paginate_records
 

src/llama_stack/providers/remote/datasetio/nvidia/datasetio.py

@@ -7,11 +7,7 @@
 from typing import Any
 
 import aiohttp
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.common.responses import PaginatedResponse
-from llama_stack.apis.common.type_system import ParamType
-from llama_stack.apis.datasets import Dataset
+from llama_stack_api import URL, Dataset, PaginatedResponse, ParamType
 
 from .config import NvidiaDatasetIOConfig
 

src/llama_stack/providers/remote/eval/nvidia/eval.py

@@ -6,18 +6,24 @@
 from typing import Any
 
 import requests
+from llama_stack_api import (
+    Agents,
+    Benchmark,
+    BenchmarkConfig,
+    BenchmarksProtocolPrivate,
+    DatasetIO,
+    Datasets,
+    Eval,
+    EvaluateResponse,
+    Inference,
+    Job,
+    JobStatus,
+    Scoring,
+    ScoringResult,
+)
 
-from llama_stack.apis.agents import Agents
-from llama_stack.apis.benchmarks import Benchmark
-from llama_stack.apis.datasetio import DatasetIO
-from llama_stack.apis.datasets import Datasets
-from llama_stack.apis.inference import Inference
-from llama_stack.apis.scoring import Scoring, ScoringResult
-from llama_stack.providers.datatypes import BenchmarksProtocolPrivate
 from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper
 
-from .....apis.common.job_types import Job, JobStatus
-from .....apis.eval.eval import BenchmarkConfig, Eval, EvaluateResponse
 from .config import NVIDIAEvalConfig
 
 DEFAULT_NAMESPACE = "nvidia"

src/llama_stack/providers/remote/files/openai/files.py

@@ -8,17 +8,17 @@ from datetime import UTC, datetime
 from typing import Annotated, Any
 
 from fastapi import Depends, File, Form, Response, UploadFile
-
-from llama_stack.apis.common.errors import ResourceNotFoundError
-from llama_stack.apis.common.responses import Order
-from llama_stack.apis.files import (
+from llama_stack_api import (
     ExpiresAfter,
     Files,
     ListOpenAIFileResponse,
     OpenAIFileDeleteResponse,
     OpenAIFileObject,
     OpenAIFilePurpose,
+    Order,
+    ResourceNotFoundError,
 )
+
 from llama_stack.core.datatypes import AccessRule
 from llama_stack.providers.utils.files.form_data import parse_expires_after
 from llama_stack.providers.utils.sqlstore.api import ColumnDefinition, ColumnType

src/llama_stack/providers/remote/files/s3/files.py

@@ -17,16 +17,17 @@ from fastapi import Depends, File, Form, Response, UploadFile
 if TYPE_CHECKING:
     from mypy_boto3_s3.client import S3Client
 
-from llama_stack.apis.common.errors import ResourceNotFoundError
-from llama_stack.apis.common.responses import Order
-from llama_stack.apis.files import (
+from llama_stack_api import (
     ExpiresAfter,
     Files,
     ListOpenAIFileResponse,
     OpenAIFileDeleteResponse,
     OpenAIFileObject,
     OpenAIFilePurpose,
+    Order,
+    ResourceNotFoundError,
 )
+
 from llama_stack.core.datatypes import AccessRule
 from llama_stack.core.id_generation import generate_object_id
 from llama_stack.providers.utils.files.form_data import parse_expires_after

src/llama_stack/providers/remote/inference/anthropic/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class AnthropicProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/azure/config.py

@@ -7,10 +7,10 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, HttpUrl, SecretStr
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class AzureProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/bedrock/bedrock.py

@@ -6,9 +6,7 @@
 
 from collections.abc import AsyncIterator, Iterable
 
-from openai import AuthenticationError
-
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
     OpenAIChatCompletionRequestWithExtraBody,
@@ -17,6 +15,8 @@ from llama_stack.apis.inference import (
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
+from openai import AuthenticationError
+
 from llama_stack.core.telemetry.tracing import get_current_span
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin

src/llama_stack/providers/remote/inference/cerebras/cerebras.py

@@ -6,10 +6,11 @@
 
 from urllib.parse import urljoin
 
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
+
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 
 from .config import CerebrasImplConfig

src/llama_stack/providers/remote/inference/cerebras/config.py

@@ -7,10 +7,10 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 DEFAULT_BASE_URL = "https://api.cerebras.ai"
 

src/llama_stack/providers/remote/inference/databricks/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class DatabricksProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/databricks/databricks.py

@@ -7,8 +7,8 @@
 from collections.abc import Iterable
 
 from databricks.sdk import WorkspaceClient
+from llama_stack_api import OpenAICompletion, OpenAICompletionRequestWithExtraBody
 
-from llama_stack.apis.inference import OpenAICompletion, OpenAICompletionRequestWithExtraBody
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 

src/llama_stack/providers/remote/inference/fireworks/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/inference/gemini/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class GeminiProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/gemini/gemini.py

@@ -6,12 +6,13 @@
 
 from typing import Any
 
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     OpenAIEmbeddingData,
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
     OpenAIEmbeddingUsage,
 )
+
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 
 from .config import GeminiConfig

src/llama_stack/providers/remote/inference/groq/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class GroqProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/llama_openai_compat/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class LlamaProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/llama_openai_compat/llama.py

@@ -4,12 +4,13 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.apis.inference.inference import (
+from llama_stack_api import (
     OpenAICompletion,
     OpenAICompletionRequestWithExtraBody,
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
+
 from llama_stack.log import get_logger
 from llama_stack.providers.remote.inference.llama_openai_compat.config import LlamaCompatConfig
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin

src/llama_stack/providers/remote/inference/nvidia/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.apis.inference import Inference
+from llama_stack_api import Inference
 
 from .config import NVIDIAConfig
 

src/llama_stack/providers/remote/inference/nvidia/config.py

@@ -7,10 +7,10 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class NVIDIAProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/nvidia/nvidia.py

@@ -8,16 +8,15 @@
 from collections.abc import Iterable
 
 import aiohttp
-
-from llama_stack.apis.inference import (
+from llama_stack_api import (
+    Model,
+    ModelType,
+    OpenAIChatCompletionContentPartImageParam,
+    OpenAIChatCompletionContentPartTextParam,
     RerankData,
     RerankResponse,
 )
-from llama_stack.apis.inference.inference import (
-    OpenAIChatCompletionContentPartImageParam,
-    OpenAIChatCompletionContentPartTextParam,
-)
-from llama_stack.apis.models import Model, ModelType
+
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 

src/llama_stack/providers/remote/inference/oci/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.apis.inference import InferenceProvider
+from llama_stack_api import InferenceProvider
 
 from .config import OCIConfig
 

src/llama_stack/providers/remote/inference/oci/config.py

@@ -7,10 +7,10 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class OCIProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/oci/oci.py

@@ -10,15 +10,15 @@ from typing import Any
 
 import httpx
 import oci
+from llama_stack_api import (
+    ModelType,
+    OpenAIEmbeddingsRequestWithExtraBody,
+    OpenAIEmbeddingsResponse,
+)
 from oci.generative_ai.generative_ai_client import GenerativeAiClient
 from oci.generative_ai.models import ModelCollection
 from openai._base_client import DefaultAsyncHttpxClient
 
-from llama_stack.apis.inference.inference import (
-    OpenAIEmbeddingsRequestWithExtraBody,
-    OpenAIEmbeddingsResponse,
-)
-from llama_stack.apis.models import ModelType
 from llama_stack.log import get_logger
 from llama_stack.providers.remote.inference.oci.auth import OciInstancePrincipalAuth, OciUserPrincipalAuth
 from llama_stack.providers.remote.inference.oci.config import OCIConfig

src/llama_stack/providers/remote/inference/ollama/ollama.py

@@ -7,15 +7,15 @@
 
 import asyncio
 
-from ollama import AsyncClient as AsyncOllamaClient
-
-from llama_stack.apis.common.errors import UnsupportedModelError
-from llama_stack.apis.models import Model
-from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import (
+from llama_stack_api import (
     HealthResponse,
     HealthStatus,
+    Model,
+    UnsupportedModelError,
 )
+from ollama import AsyncClient as AsyncOllamaClient
+
+from llama_stack.log import get_logger
 from llama_stack.providers.remote.inference.ollama.config import OllamaImplConfig
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 

src/llama_stack/providers/remote/inference/openai/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class OpenAIProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/passthrough/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/inference/passthrough/passthrough.py

@@ -6,10 +6,9 @@
 
 from collections.abc import AsyncIterator
 
-from openai import AsyncOpenAI
-
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     Inference,
+    Model,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
     OpenAIChatCompletionRequestWithExtraBody,
@@ -18,7 +17,8 @@ from llama_stack.apis.inference import (
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
-from llama_stack.apis.models import Model
+from openai import AsyncOpenAI
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
 
 from .config import PassthroughImplConfig

src/llama_stack/providers/remote/inference/runpod/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class RunpodProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/runpod/runpod.py

@@ -6,11 +6,12 @@
 
 from collections.abc import AsyncIterator
 
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
     OpenAIChatCompletionRequestWithExtraBody,
 )
+
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 
 from .config import RunpodImplConfig

src/llama_stack/providers/remote/inference/sambanova/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class SambaNovaProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/tgi/config.py

@@ -5,10 +5,10 @@
 # the root directory of this source tree.
 
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/inference/tgi/tgi.py

@@ -8,12 +8,12 @@
 from collections.abc import Iterable
 
 from huggingface_hub import AsyncInferenceClient, HfApi
-from pydantic import SecretStr
-
-from llama_stack.apis.inference import (
+from llama_stack_api import (
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
+from pydantic import SecretStr
+
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 

src/llama_stack/providers/remote/inference/together/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/inference/together/together.py

@@ -8,15 +8,15 @@
 from collections.abc import Iterable
 from typing import Any, cast
 
+from llama_stack_api import (
+    Model,
+    OpenAIEmbeddingsRequestWithExtraBody,
+    OpenAIEmbeddingsResponse,
+    OpenAIEmbeddingUsage,
+)
 from together import AsyncTogether  # type: ignore[import-untyped]
 from together.constants import BASE_URL  # type: ignore[import-untyped]
 
-from llama_stack.apis.inference import (
-    OpenAIEmbeddingsRequestWithExtraBody,
-    OpenAIEmbeddingsResponse,
-)
-from llama_stack.apis.inference.inference import OpenAIEmbeddingUsage
-from llama_stack.apis.models import Model
 from llama_stack.core.request_headers import NeedsRequestProviderData
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin

src/llama_stack/providers/remote/inference/vertexai/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class VertexAIProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/vllm/config.py

@@ -6,10 +6,10 @@
 
 from pathlib import Path
 
+from llama_stack_api import json_schema_type
 from pydantic import Field, SecretStr, field_validator
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/inference/vllm/vllm.py

@@ -7,19 +7,17 @@ from collections.abc import AsyncIterator
 from urllib.parse import urljoin
 
 import httpx
-from pydantic import ConfigDict
-
-from llama_stack.apis.inference import (
+from llama_stack_api import (
+    HealthResponse,
+    HealthStatus,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
     OpenAIChatCompletionRequestWithExtraBody,
     ToolChoice,
 )
+from pydantic import ConfigDict
+
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import (
-    HealthResponse,
-    HealthStatus,
-)
 from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
 
 from .config import VLLMInferenceAdapterConfig

src/llama_stack/providers/remote/inference/watsonx/config.py

@@ -7,10 +7,10 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.providers.utils.inference.model_registry import RemoteInferenceProviderConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 class WatsonXProviderDataValidator(BaseModel):

src/llama_stack/providers/remote/inference/watsonx/watsonx.py

@@ -9,8 +9,9 @@ from typing import Any
 
 import litellm
 import requests
-
-from llama_stack.apis.inference.inference import (
+from llama_stack_api import (
+    Model,
+    ModelType,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
     OpenAIChatCompletionRequestWithExtraBody,
@@ -20,8 +21,7 @@ from llama_stack.apis.inference.inference import (
     OpenAIEmbeddingsRequestWithExtraBody,
     OpenAIEmbeddingsResponse,
 )
-from llama_stack.apis.models import Model
-from llama_stack.apis.models.models import ModelType
+
 from llama_stack.core.telemetry.tracing import get_current_span
 from llama_stack.log import get_logger
 from llama_stack.providers.remote.inference.watsonx.config import WatsonXConfig
@@ -238,7 +238,8 @@ class WatsonXInferenceAdapter(LiteLLMOpenAIMixin):
         )
 
         # Convert response to OpenAI format
-        from llama_stack.apis.inference import OpenAIEmbeddingUsage
+        from llama_stack_api import OpenAIEmbeddingUsage
+
         from llama_stack.providers.utils.inference.litellm_openai_mixin import b64_encode_openai_embeddings_response
 
         data = b64_encode_openai_embeddings_response(response.data, params.encoding_format)

src/llama_stack/providers/remote/post_training/nvidia/README.md

@@ -128,7 +128,7 @@ client.post_training.job.cancel(job_uuid="your-job-id")
 #### 1. Register the model
 
 ```python
-from llama_stack.apis.models import Model, ModelType
+from llama_stack_api.models import Model, ModelType
 
 client.models.register(
     model_id="test-example-model@v1",

src/llama_stack/providers/remote/post_training/nvidia/post_training.py

@@ -8,9 +8,7 @@ from datetime import datetime
 from typing import Any, Literal
 
 import aiohttp
-from pydantic import BaseModel, ConfigDict
-
-from llama_stack.apis.post_training import (
+from llama_stack_api import (
     AlgorithmConfig,
     DPOAlignmentConfig,
     JobStatus,
@@ -19,6 +17,8 @@ from llama_stack.apis.post_training import (
     PostTrainingJobStatusResponse,
     TrainingConfig,
 )
+from pydantic import BaseModel, ConfigDict
+
 from llama_stack.providers.remote.post_training.nvidia.config import NvidiaPostTrainingConfig
 from llama_stack.providers.remote.post_training.nvidia.utils import warn_unsupported_params
 from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper

src/llama_stack/providers/remote/post_training/nvidia/utils.py

@@ -7,9 +7,9 @@
 import warnings
 from typing import Any
 
+from llama_stack_api import TrainingConfig
 from pydantic import BaseModel
 
-from llama_stack.apis.post_training import TrainingConfig
 from llama_stack.log import get_logger
 from llama_stack.providers.remote.post_training.nvidia.config import SFTLoRADefaultConfig
 

src/llama_stack/providers/remote/safety/bedrock/bedrock.py

@@ -7,16 +7,17 @@
 import json
 from typing import Any
 
-from llama_stack.apis.inference import OpenAIMessageParam
-from llama_stack.apis.safety import (
+from llama_stack_api import (
+    OpenAIMessageParam,
     RunShieldResponse,
     Safety,
     SafetyViolation,
+    Shield,
+    ShieldsProtocolPrivate,
     ViolationLevel,
 )
-from llama_stack.apis.shields import Shield
+
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import ShieldsProtocolPrivate
 from llama_stack.providers.utils.bedrock.client import create_bedrock_client
 
 from .config import BedrockSafetyConfig

src/llama_stack/providers/remote/safety/bedrock/config.py

@@ -5,8 +5,9 @@
 # the root directory of this source tree.
 
 
+from llama_stack_api import json_schema_type
+
 from llama_stack.providers.utils.bedrock.config import BedrockBaseConfig
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/safety/nvidia/README.md

@@ -42,8 +42,8 @@ client.initialize()
 #### Create a safety shield
 
 ```python
-from llama_stack.apis.safety import Shield
-from llama_stack.apis.inference import Message
+from llama_stack_api.safety import Shield
+from llama_stack_api.inference import Message
 
 # Create a safety shield
 shield = Shield(

src/llama_stack/providers/remote/safety/nvidia/config.py

@@ -6,10 +6,9 @@
 import os
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
-from llama_stack.schema_utils import json_schema_type
-
 
 @json_schema_type
 class NVIDIASafetyConfig(BaseModel):

src/llama_stack/providers/remote/safety/nvidia/nvidia.py

@@ -7,12 +7,18 @@
 from typing import Any
 
 import requests
+from llama_stack_api import (
+    ModerationObject,
+    OpenAIMessageParam,
+    RunShieldResponse,
+    Safety,
+    SafetyViolation,
+    Shield,
+    ShieldsProtocolPrivate,
+    ViolationLevel,
+)
 
-from llama_stack.apis.inference import OpenAIMessageParam
-from llama_stack.apis.safety import ModerationObject, RunShieldResponse, Safety, SafetyViolation, ViolationLevel
-from llama_stack.apis.shields import Shield
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import ShieldsProtocolPrivate
 
 from .config import NVIDIASafetyConfig
 

src/llama_stack/providers/remote/safety/sambanova/config.py

@@ -6,10 +6,9 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
 
-from llama_stack.schema_utils import json_schema_type
-
 
 class SambaNovaProviderDataValidator(BaseModel):
     sambanova_api_key: str | None = Field(

src/llama_stack/providers/remote/safety/sambanova/sambanova.py

@@ -8,18 +8,18 @@ from typing import Any
 
 import litellm
 import requests
-
-from llama_stack.apis.inference import OpenAIMessageParam
-from llama_stack.apis.safety import (
+from llama_stack_api import (
+    OpenAIMessageParam,
     RunShieldResponse,
     Safety,
     SafetyViolation,
+    Shield,
+    ShieldsProtocolPrivate,
     ViolationLevel,
 )
-from llama_stack.apis.shields import Shield
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import ShieldsProtocolPrivate
 
 from .config import SambaNovaSafetyConfig
 

src/llama_stack/providers/remote/tool_runtime/bing_search/bing_search.py

@@ -8,17 +8,17 @@ import json
 from typing import Any
 
 import httpx
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.tools import (
+from llama_stack_api import (
+    URL,
     ListToolDefsResponse,
     ToolDef,
     ToolGroup,
+    ToolGroupsProtocolPrivate,
     ToolInvocationResult,
     ToolRuntime,
 )
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
-from llama_stack.providers.datatypes import ToolGroupsProtocolPrivate
 
 from .config import BingSearchToolConfig
 

src/llama_stack/providers/remote/tool_runtime/brave_search/brave_search.py

@@ -7,18 +7,18 @@
 from typing import Any
 
 import httpx
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.tools import (
+from llama_stack_api import (
+    URL,
     ListToolDefsResponse,
     ToolDef,
     ToolGroup,
+    ToolGroupsProtocolPrivate,
     ToolInvocationResult,
     ToolRuntime,
 )
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
 from llama_stack.models.llama.datatypes import BuiltinTool
-from llama_stack.providers.datatypes import ToolGroupsProtocolPrivate
 
 from .config import BraveSearchToolConfig
 

src/llama_stack/providers/remote/tool_runtime/model_context_protocol/model_context_protocol.py

@@ -7,17 +7,18 @@
 from typing import Any
 from urllib.parse import urlparse
 
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.datatypes import Api
-from llama_stack.apis.tools import (
+from llama_stack_api import (
+    URL,
+    Api,
     ListToolDefsResponse,
     ToolGroup,
+    ToolGroupsProtocolPrivate,
     ToolInvocationResult,
     ToolRuntime,
 )
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import ToolGroupsProtocolPrivate
 from llama_stack.providers.utils.tools.mcp import invoke_mcp_tool, list_mcp_tools
 
 from .config import MCPProviderConfig

src/llama_stack/providers/remote/tool_runtime/tavily_search/tavily_search.py

@@ -8,17 +8,17 @@ import json
 from typing import Any
 
 import httpx
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.tools import (
+from llama_stack_api import (
+    URL,
     ListToolDefsResponse,
     ToolDef,
     ToolGroup,
+    ToolGroupsProtocolPrivate,
     ToolInvocationResult,
     ToolRuntime,
 )
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
-from llama_stack.providers.datatypes import ToolGroupsProtocolPrivate
 
 from .config import TavilySearchToolConfig
 

src/llama_stack/providers/remote/tool_runtime/wolfram_alpha/wolfram_alpha.py

@@ -8,17 +8,17 @@ import json
 from typing import Any
 
 import httpx
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.tools import (
+from llama_stack_api import (
+    URL,
     ListToolDefsResponse,
     ToolDef,
     ToolGroup,
+    ToolGroupsProtocolPrivate,
     ToolInvocationResult,
     ToolRuntime,
 )
+
 from llama_stack.core.request_headers import NeedsRequestProviderData
-from llama_stack.providers.datatypes import ToolGroupsProtocolPrivate
 
 from .config import WolframAlphaToolConfig
 

src/llama_stack/providers/remote/vector_io/chroma/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, ProviderSpec
+from llama_stack_api import Api, ProviderSpec
 
 from .config import ChromaVectorIOConfig
 

src/llama_stack/providers/remote/vector_io/chroma/chroma.py

@@ -9,14 +9,19 @@ from typing import Any
 from urllib.parse import urlparse
 
 import chromadb
+from llama_stack_api import (
+    Chunk,
+    Files,
+    Inference,
+    InterleavedContent,
+    QueryChunksResponse,
+    VectorIO,
+    VectorStore,
+    VectorStoresProtocolPrivate,
+)
 from numpy.typing import NDArray
 
-from llama_stack.apis.files import Files
-from llama_stack.apis.inference import Inference, InterleavedContent
-from llama_stack.apis.vector_io import Chunk, QueryChunksResponse, VectorIO
-from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import VectorStoresProtocolPrivate
 from llama_stack.providers.inline.vector_io.chroma import ChromaVectorIOConfig as InlineChromaVectorIOConfig
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.kvstore.api import KVStore

src/llama_stack/providers/remote/vector_io/chroma/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.core.storage.datatypes import KVStoreReference
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/vector_io/milvus/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, ProviderSpec
+from llama_stack_api import Api, ProviderSpec
 
 from .config import MilvusVectorIOConfig
 

src/llama_stack/providers/remote/vector_io/milvus/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, ConfigDict, Field
 
 from llama_stack.core.storage.datatypes import KVStoreReference
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/vector_io/milvus/milvus.py

@@ -8,16 +8,21 @@ import asyncio
 import os
 from typing import Any
 
+from llama_stack_api import (
+    Chunk,
+    Files,
+    Inference,
+    InterleavedContent,
+    QueryChunksResponse,
+    VectorIO,
+    VectorStore,
+    VectorStoreNotFoundError,
+    VectorStoresProtocolPrivate,
+)
 from numpy.typing import NDArray
 from pymilvus import AnnSearchRequest, DataType, Function, FunctionType, MilvusClient, RRFRanker, WeightedRanker
 
-from llama_stack.apis.common.errors import VectorStoreNotFoundError
-from llama_stack.apis.files import Files
-from llama_stack.apis.inference import Inference, InterleavedContent
-from llama_stack.apis.vector_io import Chunk, QueryChunksResponse, VectorIO
-from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import VectorStoresProtocolPrivate
 from llama_stack.providers.inline.vector_io.milvus import MilvusVectorIOConfig as InlineMilvusVectorIOConfig
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.kvstore.api import KVStore

src/llama_stack/providers/remote/vector_io/pgvector/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, ProviderSpec
+from llama_stack_api import Api, ProviderSpec
 
 from .config import PGVectorVectorIOConfig
 

src/llama_stack/providers/remote/vector_io/pgvector/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.core.storage.datatypes import KVStoreReference
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/vector_io/pgvector/pgvector.py

@@ -8,18 +8,23 @@ import heapq
 from typing import Any
 
 import psycopg2
+from llama_stack_api import (
+    Chunk,
+    Files,
+    Inference,
+    InterleavedContent,
+    QueryChunksResponse,
+    VectorIO,
+    VectorStore,
+    VectorStoreNotFoundError,
+    VectorStoresProtocolPrivate,
+)
 from numpy.typing import NDArray
 from psycopg2 import sql
 from psycopg2.extras import Json, execute_values
 from pydantic import BaseModel, TypeAdapter
 
-from llama_stack.apis.common.errors import VectorStoreNotFoundError
-from llama_stack.apis.files import Files
-from llama_stack.apis.inference import Inference, InterleavedContent
-from llama_stack.apis.vector_io import Chunk, QueryChunksResponse, VectorIO
-from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import VectorStoresProtocolPrivate
 from llama_stack.providers.utils.inference.prompt_adapter import interleaved_content_as_str
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.kvstore.api import KVStore

src/llama_stack/providers/remote/vector_io/qdrant/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, ProviderSpec
+from llama_stack_api import Api, ProviderSpec
 
 from .config import QdrantVectorIOConfig
 

src/llama_stack/providers/remote/vector_io/qdrant/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel
 
 from llama_stack.core.storage.datatypes import KVStoreReference
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/vector_io/qdrant/qdrant.py

@@ -9,23 +9,24 @@ import hashlib
 import uuid
 from typing import Any
 
+from llama_stack_api import (
+    Chunk,
+    Files,
+    Inference,
+    InterleavedContent,
+    QueryChunksResponse,
+    VectorIO,
+    VectorStore,
+    VectorStoreChunkingStrategy,
+    VectorStoreFileObject,
+    VectorStoreNotFoundError,
+    VectorStoresProtocolPrivate,
+)
 from numpy.typing import NDArray
 from qdrant_client import AsyncQdrantClient, models
 from qdrant_client.models import PointStruct
 
-from llama_stack.apis.common.errors import VectorStoreNotFoundError
-from llama_stack.apis.files import Files
-from llama_stack.apis.inference import Inference, InterleavedContent
-from llama_stack.apis.vector_io import (
-    Chunk,
-    QueryChunksResponse,
-    VectorIO,
-    VectorStoreChunkingStrategy,
-    VectorStoreFileObject,
-)
-from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import VectorStoresProtocolPrivate
 from llama_stack.providers.inline.vector_io.qdrant import QdrantVectorIOConfig as InlineQdrantVectorIOConfig
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.memory.openai_vector_store_mixin import OpenAIVectorStoreMixin

src/llama_stack/providers/remote/vector_io/weaviate/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, ProviderSpec
+from llama_stack_api import Api, ProviderSpec
 
 from .config import WeaviateVectorIOConfig
 

src/llama_stack/providers/remote/vector_io/weaviate/config.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
+from llama_stack_api import json_schema_type
 from pydantic import BaseModel, Field
 
 from llama_stack.core.storage.datatypes import KVStoreReference
-from llama_stack.schema_utils import json_schema_type
 
 
 @json_schema_type

src/llama_stack/providers/remote/vector_io/weaviate/weaviate.py

@@ -8,19 +8,23 @@ from typing import Any
 
 import weaviate
 import weaviate.classes as wvc
+from llama_stack_api import (
+    Chunk,
+    Files,
+    Inference,
+    InterleavedContent,
+    QueryChunksResponse,
+    VectorIO,
+    VectorStore,
+    VectorStoreNotFoundError,
+    VectorStoresProtocolPrivate,
+)
 from numpy.typing import NDArray
 from weaviate.classes.init import Auth
 from weaviate.classes.query import Filter, HybridFusion
 
-from llama_stack.apis.common.content_types import InterleavedContent
-from llama_stack.apis.common.errors import VectorStoreNotFoundError
-from llama_stack.apis.files import Files
-from llama_stack.apis.inference import Inference
-from llama_stack.apis.vector_io import Chunk, QueryChunksResponse, VectorIO
-from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.core.request_headers import NeedsRequestProviderData
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import VectorStoresProtocolPrivate
 from llama_stack.providers.utils.kvstore import kvstore_impl
 from llama_stack.providers.utils.kvstore.api import KVStore
 from llama_stack.providers.utils.memory.openai_vector_store_mixin import OpenAIVectorStoreMixin