feat: refactor llama-stack-api structure

move llama_stack_api.apis... to top level llama_stack_api.

merge provider datatypes and the existing apis.datatypes into a common llama_stack_api.datatypes

update all usages of these packages throughout LLS

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

docs/docs/providers/vector_io/inline_sqlite-vec.mdx

@@ -153,7 +153,7 @@ description: |
   Example using RAGQueryConfig with different search modes:
 
   ```python
-  from llama_stack_api.apis.tools import RAGQueryConfig, RRFRanker, WeightedRanker
+  from llama_stack_api.rag_tool import RAGQueryConfig, RRFRanker, WeightedRanker
 
   # Vector search
   config = RAGQueryConfig(mode="vector", max_chunks=5)
@@ -358,7 +358,7 @@ Two ranker types are supported:
 Example using RAGQueryConfig with different search modes:
 
 ```python
-from llama_stack_api.apis.tools import RAGQueryConfig, RRFRanker, WeightedRanker
+from llama_stack_api.rag_tool import RAGQueryConfig, RRFRanker, WeightedRanker
 
 # Vector search
 config = RAGQueryConfig(mode="vector", max_chunks=5)

docs/openapi_generator/generate.py

@@ -16,7 +16,7 @@ import sys
 import fire
 import ruamel.yaml as yaml
 
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1 # noqa: E402
+from llama_stack_api.version import LLAMA_STACK_API_V1 # noqa: E402
 from llama_stack.core.stack import LlamaStack  # noqa: E402
 
 from .pyopenapi.options import Options  # noqa: E402

docs/openapi_generator/pyopenapi/generator.py

@@ -16,7 +16,7 @@ from typing import Annotated, Any, Dict, get_args, get_origin, Set, Union
 
 from fastapi import UploadFile
 
-from llama_stack_api.apis.datatypes import Error
+from llama_stack_api.datatypes import Error
 from llama_stack_api.strong_typing.core import JsonType
 from llama_stack_api.strong_typing.docstring import Docstring, parse_type
 from llama_stack_api.strong_typing.inspection import (

docs/openapi_generator/pyopenapi/operations.py

@@ -11,7 +11,7 @@ import typing
 from dataclasses import dataclass
 from typing import Any, Callable, Dict, Iterable, Iterator, List, Optional, Tuple, Union
 
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1BETA, LLAMA_STACK_API_V1ALPHA
+from llama_stack_api.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1BETA, LLAMA_STACK_API_V1ALPHA
 
 from termcolor import colored
 

scripts/generate_prompt_format.py

@@ -14,7 +14,7 @@ import os
 from pathlib import Path
 
 import fire
-from llama_stack_api.apis.common.errors import ModelNotFoundError
+from llama_stack_api.common.errors import ModelNotFoundError
 
 from llama_stack.models.llama.llama3.generation import Llama3
 from llama_stack.models.llama.llama4.generation import Llama4

scripts/provider_codegen.py

@@ -22,7 +22,7 @@ def get_api_docstring(api_name: str) -> str | None:
     """Extract docstring from the API protocol class."""
     try:
         # Import the API module dynamically
-        api_module = __import__(f"llama_stack_api.apis.{api_name}", fromlist=[api_name.title()])
+        api_module = __import__(f"llama_stack_api.{api_name}", fromlist=[api_name.title()])
 
         # Get the main protocol class (usually capitalized API name)
         protocol_class_name = api_name.title()
@@ -83,9 +83,9 @@ def get_config_class_info(config_class_path: str) -> dict[str, Any]:
                 # this string replace is ridiculous
                 field_type = field_type.replace("typing.", "").replace("Optional[", "").replace("]", "")
                 field_type = field_type.replace("Annotated[", "").replace("FieldInfo(", "").replace(")", "")
-                field_type = field_type.replace("llama_stack_api.apis.inference.inference.", "")
+                field_type = field_type.replace("llama_stack_api.inference.", "")
                 field_type = field_type.replace("llama_stack.providers.", "")
-                field_type = field_type.replace("llama_stack_api.providers.", "")
+                field_type = field_type.replace("llama_stack_api.datatypes.", "")
 
                 default_value = field.default
                 if field.default_factory is not None:

src/llama-stack-api/README.md

@@ -58,9 +58,9 @@ Current version: **0.1.0**
 ## Usage Example
 
 ```python
-from llama_stack_api.apis.inference import Inference, ChatCompletionRequest
+from llama_stack_api.inference import Inference, ChatCompletionRequest
 from llama_stack_api.providers.datatypes import ProviderSpec, InlineProviderSpec
-from llama_stack_api.apis.datatypes import Api
+from llama_stack_api.datatypes import Api
 
 
 # Use protocol definitions for type checking

src/llama-stack-api/llama_stack_api/__init__.py

@@ -13,14 +13,15 @@ and clients that need to interact with Llama Stack APIs without requiring the fu
 server implementation.
 
 Key components:
-- apis: Protocol definitions for all Llama Stack APIs
-- providers: Provider interface specifications
+- API modules (agents, inference, safety, etc.): Protocol definitions for all Llama Stack APIs
+- datatypes: Core data types and provider specifications
+- common: Common data types used across APIs
 - strong_typing: Type system utilities
 - schema_utils: Schema validation and utilities
 """
 
 __version__ = "0.1.0"
 
-from . import apis, providers, schema_utils, strong_typing  # noqa: F401
+from . import common, datatypes, schema_utils, strong_typing  # noqa: F401
 
-__all__ = ["apis", "providers", "schema_utils", "strong_typing"]
+__all__ = ["common", "datatypes", "schema_utils", "strong_typing"]

src/llama-stack-api/llama_stack_api/agents.py

@@ -9,9 +9,9 @@ from typing import Annotated, Protocol, runtime_checkable
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.common.responses import Order
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.responses import Order
 from llama_stack_api.schema_utils import ExtraBodyField, json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 from .openai_responses import (
     ListOpenAIResponseInputItem,

src/llama-stack-api/llama_stack_api/apis/agents/__init__.py

@@ -1,7 +0,0 @@
-# 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 .agents import *

src/llama-stack-api/llama_stack_api/apis/batches/__init__.py

@@ -1,9 +0,0 @@
-# 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 .batches import Batches, BatchObject, ListBatchesResponse
-
-__all__ = ["Batches", "BatchObject", "ListBatchesResponse"]

src/llama-stack-api/llama_stack_api/apis/benchmarks/__init__.py

@@ -1,7 +0,0 @@
-# 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 .benchmarks import *

src/llama-stack-api/llama_stack_api/apis/common/__init__.py

@@ -1,5 +0,0 @@
-# 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.

src/llama-stack-api/llama_stack_api/apis/conversations/__init__.py

@@ -1,27 +0,0 @@
-# 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 .conversations import (
-    Conversation,
-    ConversationDeletedResource,
-    ConversationItem,
-    ConversationItemCreateRequest,
-    ConversationItemDeletedResource,
-    ConversationItemList,
-    Conversations,
-    Metadata,
-)
-
-__all__ = [
-    "Conversation",
-    "ConversationDeletedResource",
-    "ConversationItem",
-    "ConversationItemCreateRequest",
-    "ConversationItemDeletedResource",
-    "ConversationItemList",
-    "Conversations",
-    "Metadata",
-]

src/llama-stack-api/llama_stack_api/apis/datasetio/__init__.py

@@ -1,7 +0,0 @@
-# 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 .datasetio import *

src/llama-stack-api/llama_stack_api/apis/datasets/__init__.py

@@ -1,7 +0,0 @@
-# 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 .datasets import *

src/llama-stack-api/llama_stack_api/apis/datatypes.py

@@ -1,158 +0,0 @@
-# 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 enum import Enum, EnumMeta
-
-from pydantic import BaseModel, Field
-
-from llama_stack_api.schema_utils import json_schema_type
-
-
-class DynamicApiMeta(EnumMeta):
-    def __new__(cls, name, bases, namespace):
-        # Store the original enum values
-        original_values = {k: v for k, v in namespace.items() if not k.startswith("_")}
-
-        # Create the enum class
-        cls = super().__new__(cls, name, bases, namespace)
-
-        # Store the original values for reference
-        cls._original_values = original_values
-        # Initialize _dynamic_values
-        cls._dynamic_values = {}
-
-        return cls
-
-    def __call__(cls, value):
-        try:
-            return super().__call__(value)
-        except ValueError as e:
-            # If this value was already dynamically added, return it
-            if value in cls._dynamic_values:
-                return cls._dynamic_values[value]
-
-            # If the value doesn't exist, create a new enum member
-            # Create a new member name from the value
-            member_name = value.lower().replace("-", "_")
-
-            # If this member name already exists in the enum, return the existing member
-            if member_name in cls._member_map_:
-                return cls._member_map_[member_name]
-
-            # Instead of creating a new member, raise ValueError to force users to use Api.add() to
-            # register new APIs explicitly
-            raise ValueError(f"API '{value}' does not exist. Use Api.add() to register new APIs.") from e
-
-    def __iter__(cls):
-        # Allow iteration over both static and dynamic members
-        yield from super().__iter__()
-        if hasattr(cls, "_dynamic_values"):
-            yield from cls._dynamic_values.values()
-
-    def add(cls, value):
-        """
-        Add a new API to the enum.
-        Used to register external APIs.
-        """
-        member_name = value.lower().replace("-", "_")
-
-        # If this member name already exists in the enum, return it
-        if member_name in cls._member_map_:
-            return cls._member_map_[member_name]
-
-        # Create a new enum member
-        member = object.__new__(cls)
-        member._name_ = member_name
-        member._value_ = value
-
-        # Add it to the enum class
-        cls._member_map_[member_name] = member
-        cls._member_names_.append(member_name)
-        cls._member_type_ = str
-
-        # Store it in our dynamic values
-        cls._dynamic_values[value] = member
-
-        return member
-
-
-@json_schema_type
-class Api(Enum, metaclass=DynamicApiMeta):
-    """Enumeration of all available APIs in the Llama Stack system.
-    :cvar providers: Provider management and configuration
-    :cvar inference: Text generation, chat completions, and embeddings
-    :cvar safety: Content moderation and safety shields
-    :cvar agents: Agent orchestration and execution
-    :cvar batches: Batch processing for asynchronous API requests
-    :cvar vector_io: Vector database operations and queries
-    :cvar datasetio: Dataset input/output operations
-    :cvar scoring: Model output evaluation and scoring
-    :cvar eval: Model evaluation and benchmarking framework
-    :cvar post_training: Fine-tuning and model training
-    :cvar tool_runtime: Tool execution and management
-    :cvar telemetry: Observability and system monitoring
-    :cvar models: Model metadata and management
-    :cvar shields: Safety shield implementations
-    :cvar datasets: Dataset creation and management
-    :cvar scoring_functions: Scoring function definitions
-    :cvar benchmarks: Benchmark suite management
-    :cvar tool_groups: Tool group organization
-    :cvar files: File storage and management
-    :cvar prompts: Prompt versions and management
-    :cvar inspect: Built-in system inspection and introspection
-    """
-
-    providers = "providers"
-    inference = "inference"
-    safety = "safety"
-    agents = "agents"
-    batches = "batches"
-    vector_io = "vector_io"
-    datasetio = "datasetio"
-    scoring = "scoring"
-    eval = "eval"
-    post_training = "post_training"
-    tool_runtime = "tool_runtime"
-
-    models = "models"
-    shields = "shields"
-    vector_stores = "vector_stores"  # only used for routing table
-    datasets = "datasets"
-    scoring_functions = "scoring_functions"
-    benchmarks = "benchmarks"
-    tool_groups = "tool_groups"
-    files = "files"
-    prompts = "prompts"
-    conversations = "conversations"
-
-    # built-in API
-    inspect = "inspect"
-
-
-@json_schema_type
-class Error(BaseModel):
-    """
-    Error response from the API. Roughly follows RFC 7807.
-
-    :param status: HTTP status code
-    :param title: Error title, a short summary of the error which is invariant for an error type
-    :param detail: Error detail, a longer human-readable description of the error
-    :param instance: (Optional) A URL which can be used to retrieve more information about the specific occurrence of the error
-    """
-
-    status: int
-    title: str
-    detail: str
-    instance: str | None = None
-
-
-class ExternalApiSpec(BaseModel):
-    """Specification for an external API implementation."""
-
-    module: str = Field(..., description="Python module containing the API implementation")
-    name: str = Field(..., description="Name of the API")
-    pip_packages: list[str] = Field(default=[], description="List of pip packages to install the API")
-    protocol: str = Field(..., description="Name of the protocol class for the API")

src/llama-stack-api/llama_stack_api/apis/eval/__init__.py

@@ -1,7 +0,0 @@
-# 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 .eval import *

src/llama-stack-api/llama_stack_api/apis/files/__init__.py

@@ -1,7 +0,0 @@
-# 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 .files import *

src/llama-stack-api/llama_stack_api/apis/inference/__init__.py

@@ -1,7 +0,0 @@
-# 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 .inference import *

src/llama-stack-api/llama_stack_api/apis/inspect/__init__.py

@@ -1,7 +0,0 @@
-# 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 .inspect import *

src/llama-stack-api/llama_stack_api/apis/models/__init__.py

@@ -1,7 +0,0 @@
-# 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 .models import *

src/llama-stack-api/llama_stack_api/apis/post_training/__init__.py

@@ -1,7 +0,0 @@
-# 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 .post_training import *

src/llama-stack-api/llama_stack_api/apis/prompts/__init__.py

@@ -1,9 +0,0 @@
-# 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 .prompts import ListPromptsResponse, Prompt, Prompts
-
-__all__ = ["Prompt", "Prompts", "ListPromptsResponse"]

src/llama-stack-api/llama_stack_api/apis/providers/__init__.py

@@ -1,7 +0,0 @@
-# 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 .providers import *

src/llama-stack-api/llama_stack_api/apis/safety/__init__.py

@@ -1,7 +0,0 @@
-# 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 .safety import *

src/llama-stack-api/llama_stack_api/apis/scoring/__init__.py

@@ -1,7 +0,0 @@
-# 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 .scoring import *

src/llama-stack-api/llama_stack_api/apis/scoring_functions/__init__.py

@@ -1,7 +0,0 @@
-# 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 .scoring_functions import *

src/llama-stack-api/llama_stack_api/apis/shields/__init__.py

@@ -1,7 +0,0 @@
-# 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 .shields import *

src/llama-stack-api/llama_stack_api/apis/tools/__init__.py

@@ -1,8 +0,0 @@
-# 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 .rag_tool import *
-from .tools import *

src/llama-stack-api/llama_stack_api/apis/vector_io/__init__.py

@@ -1,7 +0,0 @@
-# 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 .vector_io import *

src/llama-stack-api/llama_stack_api/apis/vector_stores/__init__.py

@@ -1,7 +0,0 @@
-# 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 .vector_stores import *

src/llama-stack-api/llama_stack_api/batches.py

@@ -8,8 +8,8 @@ from typing import Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 try:
     from openai.types import Batch as BatchObject

src/llama-stack-api/llama_stack_api/benchmarks.py

@@ -7,9 +7,9 @@ from typing import Any, Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1ALPHA
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1ALPHA
 
 
 class CommonBenchmarkFields(BaseModel):

src/llama-stack-api/llama_stack_api/conversations.py

@@ -9,7 +9,8 @@ from typing import Annotated, Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.agents.openai_responses import (
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.openai_responses import (
     OpenAIResponseInputFunctionToolCallOutput,
     OpenAIResponseMCPApprovalRequest,
     OpenAIResponseMCPApprovalResponse,
@@ -20,9 +21,8 @@ from llama_stack_api.apis.agents.openai_responses import (
     OpenAIResponseOutputMessageMCPListTools,
     OpenAIResponseOutputMessageWebSearchToolCall,
 )
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
 from llama_stack_api.schema_utils import json_schema_type, register_schema, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 Metadata = dict[str, str]
 

src/llama-stack-api/llama_stack_api/datasetio.py

@@ -6,10 +6,10 @@
 
 from typing import Any, Protocol, runtime_checkable
 
-from llama_stack_api.apis.common.responses import PaginatedResponse
-from llama_stack_api.apis.datasets import Dataset
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1BETA
+from llama_stack_api.common.responses import PaginatedResponse
+from llama_stack_api.datasets import Dataset
 from llama_stack_api.schema_utils import webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1BETA
 
 
 class DatasetStore(Protocol):

src/llama-stack-api/llama_stack_api/datasets.py

@@ -9,9 +9,9 @@ from typing import Annotated, Any, Literal, Protocol
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1BETA
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, register_schema, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1BETA
 
 
 class DatasetPurpose(StrEnum):

src/llama-stack-api/llama_stack_api/datatypes.py

@@ -4,21 +4,172 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from enum import StrEnum
+from enum import Enum, EnumMeta, StrEnum
 from typing import Any, Protocol
 from urllib.parse import urlparse
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.benchmarks import Benchmark
-from llama_stack_api.apis.datasets import Dataset
-from llama_stack_api.apis.datatypes import Api
-from llama_stack_api.apis.models import Model
-from llama_stack_api.apis.scoring_functions import ScoringFn
-from llama_stack_api.apis.shields import Shield
-from llama_stack_api.apis.tools import ToolGroup
-from llama_stack_api.apis.vector_stores import VectorStore
+from llama_stack_api.benchmarks import Benchmark
+from llama_stack_api.datasets import Dataset
+from llama_stack_api.models import Model
 from llama_stack_api.schema_utils import json_schema_type
+from llama_stack_api.scoring_functions import ScoringFn
+from llama_stack_api.shields import Shield
+from llama_stack_api.tools import ToolGroup
+from llama_stack_api.vector_stores import VectorStore
+
+
+class DynamicApiMeta(EnumMeta):
+    def __new__(cls, name, bases, namespace):
+        # Store the original enum values
+        original_values = {k: v for k, v in namespace.items() if not k.startswith("_")}
+
+        # Create the enum class
+        cls = super().__new__(cls, name, bases, namespace)
+
+        # Store the original values for reference
+        cls._original_values = original_values
+        # Initialize _dynamic_values
+        cls._dynamic_values = {}
+
+        return cls
+
+    def __call__(cls, value):
+        try:
+            return super().__call__(value)
+        except ValueError as e:
+            # If this value was already dynamically added, return it
+            if value in cls._dynamic_values:
+                return cls._dynamic_values[value]
+
+            # If the value doesn't exist, create a new enum member
+            # Create a new member name from the value
+            member_name = value.lower().replace("-", "_")
+
+            # If this member name already exists in the enum, return the existing member
+            if member_name in cls._member_map_:
+                return cls._member_map_[member_name]
+
+            # Instead of creating a new member, raise ValueError to force users to use Api.add() to
+            # register new APIs explicitly
+            raise ValueError(f"API '{value}' does not exist. Use Api.add() to register new APIs.") from e
+
+    def __iter__(cls):
+        # Allow iteration over both static and dynamic members
+        yield from super().__iter__()
+        if hasattr(cls, "_dynamic_values"):
+            yield from cls._dynamic_values.values()
+
+    def add(cls, value):
+        """
+        Add a new API to the enum.
+        Used to register external APIs.
+        """
+        member_name = value.lower().replace("-", "_")
+
+        # If this member name already exists in the enum, return it
+        if member_name in cls._member_map_:
+            return cls._member_map_[member_name]
+
+        # Create a new enum member
+        member = object.__new__(cls)
+        member._name_ = member_name
+        member._value_ = value
+
+        # Add it to the enum class
+        cls._member_map_[member_name] = member
+        cls._member_names_.append(member_name)
+        cls._member_type_ = str
+
+        # Store it in our dynamic values
+        cls._dynamic_values[value] = member
+
+        return member
+
+
+@json_schema_type
+class Api(Enum, metaclass=DynamicApiMeta):
+    """Enumeration of all available APIs in the Llama Stack system.
+    :cvar providers: Provider management and configuration
+    :cvar inference: Text generation, chat completions, and embeddings
+    :cvar safety: Content moderation and safety shields
+    :cvar agents: Agent orchestration and execution
+    :cvar batches: Batch processing for asynchronous API requests
+    :cvar vector_io: Vector database operations and queries
+    :cvar datasetio: Dataset input/output operations
+    :cvar scoring: Model output evaluation and scoring
+    :cvar eval: Model evaluation and benchmarking framework
+    :cvar post_training: Fine-tuning and model training
+    :cvar tool_runtime: Tool execution and management
+    :cvar telemetry: Observability and system monitoring
+    :cvar models: Model metadata and management
+    :cvar shields: Safety shield implementations
+    :cvar datasets: Dataset creation and management
+    :cvar scoring_functions: Scoring function definitions
+    :cvar benchmarks: Benchmark suite management
+    :cvar tool_groups: Tool group organization
+    :cvar files: File storage and management
+    :cvar prompts: Prompt versions and management
+    :cvar inspect: Built-in system inspection and introspection
+    """
+
+    providers = "providers"
+    inference = "inference"
+    safety = "safety"
+    agents = "agents"
+    batches = "batches"
+    vector_io = "vector_io"
+    datasetio = "datasetio"
+    scoring = "scoring"
+    eval = "eval"
+    post_training = "post_training"
+    tool_runtime = "tool_runtime"
+
+    models = "models"
+    shields = "shields"
+    vector_stores = "vector_stores"  # only used for routing table
+    datasets = "datasets"
+    scoring_functions = "scoring_functions"
+    benchmarks = "benchmarks"
+    tool_groups = "tool_groups"
+    files = "files"
+    prompts = "prompts"
+    conversations = "conversations"
+
+    # built-in API
+    inspect = "inspect"
+
+
+@json_schema_type
+class Error(BaseModel):
+    """
+    Error response from the API. Roughly follows RFC 7807.
+
+    :param status: HTTP status code
+    :param title: Error title, a short summary of the error which is invariant for an error type
+    :param detail: Error detail, a longer human-readable description of the error
+    :param instance: (Optional) A URL which can be used to retrieve more information about the specific occurrence of the error
+    """
+
+    status: int
+    title: str
+    detail: str
+    instance: str | None = None
+
+
+class ExternalApiSpec(BaseModel):
+    """Specification for an external API implementation."""
+
+    module: str = Field(..., description="Python module containing the API implementation")
+    name: str = Field(..., description="Name of the API")
+    pip_packages: list[str] = Field(default=[], description="List of pip packages to install the API")
+    protocol: str = Field(..., description="Name of the protocol class for the API")
+
+
+# Provider-related types (merged from providers/datatypes.py)
+# NOTE: These imports are forward references to avoid circular dependencies
+# They will be resolved at runtime when the classes are used
 
 
 class ModelsProtocolPrivate(Protocol):

src/llama-stack-api/llama_stack_api/eval.py

@@ -8,12 +8,12 @@ from typing import Any, Literal, Protocol
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.job_types import Job
-from llama_stack_api.apis.inference import SamplingParams, SystemMessage
-from llama_stack_api.apis.scoring import ScoringResult
-from llama_stack_api.apis.scoring_functions import ScoringFnParams
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1ALPHA
+from llama_stack_api.common.job_types import Job
+from llama_stack_api.inference import SamplingParams, SystemMessage
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.scoring import ScoringResult
+from llama_stack_api.scoring_functions import ScoringFnParams
+from llama_stack_api.version import LLAMA_STACK_API_V1ALPHA
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/files.py

@@ -10,10 +10,10 @@ from typing import Annotated, ClassVar, Literal, Protocol, runtime_checkable
 from fastapi import File, Form, Response, UploadFile
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.responses import Order
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.responses import Order
+from llama_stack_api.common.tracing import telemetry_traceable
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 # OpenAI Files API Models

src/llama-stack-api/llama_stack_api/inference.py

@@ -18,14 +18,14 @@ from fastapi import Body
 from pydantic import BaseModel, Field
 from typing_extensions import TypedDict
 
-from llama_stack_api.apis.common.content_types import InterleavedContent
-from llama_stack_api.apis.common.responses import (
+from llama_stack_api.common.content_types import InterleavedContent
+from llama_stack_api.common.responses import (
     Order,
 )
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.models import Model
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1ALPHA
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.models import Model
 from llama_stack_api.schema_utils import json_schema_type, register_schema, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1ALPHA
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/inspect.py

@@ -8,11 +8,11 @@ from typing import Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.version import (
+from llama_stack_api.datatypes import HealthStatus
+from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import (
     LLAMA_STACK_API_V1,
 )
-from llama_stack_api.providers.datatypes import HealthStatus
-from llama_stack_api.schema_utils import json_schema_type, webmethod
 
 # Valid values for the route filter parameter.
 # Actual API levels: v1, v1alpha, v1beta (filters by level, excludes deprecated)

src/llama-stack-api/llama_stack_api/models.py

@@ -9,10 +9,10 @@ from typing import Any, Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel, ConfigDict, Field, field_validator
 
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 class CommonModelFields(BaseModel):

src/llama-stack-api/llama_stack_api/openai_responses.py

@@ -10,8 +10,8 @@ from typing import Annotated, Any, Literal
 from pydantic import BaseModel, Field, model_validator
 from typing_extensions import TypedDict
 
-from llama_stack_api.apis.vector_io import SearchRankingOptions as FileSearchRankingOptions
 from llama_stack_api.schema_utils import json_schema_type, register_schema
+from llama_stack_api.vector_io import SearchRankingOptions as FileSearchRankingOptions
 
 # NOTE(ashwin): this file is literally a copy of the OpenAI responses API schema. We should probably
 # take their YAML and generate this file automatically. Their YAML is available.

src/llama-stack-api/llama_stack_api/post_training.py

@@ -10,11 +10,11 @@ from typing import Annotated, Any, Literal, Protocol
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.content_types import URL
-from llama_stack_api.apis.common.job_types import JobStatus
-from llama_stack_api.apis.common.training_types import Checkpoint
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1ALPHA
+from llama_stack_api.common.content_types import URL
+from llama_stack_api.common.job_types import JobStatus
+from llama_stack_api.common.training_types import Checkpoint
 from llama_stack_api.schema_utils import json_schema_type, register_schema, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1ALPHA
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/prompts.py

@@ -10,9 +10,9 @@ from typing import Protocol, runtime_checkable
 
 from pydantic import BaseModel, Field, field_validator, model_validator
 
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.tracing import telemetry_traceable
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/providers.py

@@ -8,9 +8,9 @@ from typing import Any, Protocol, runtime_checkable
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
-from llama_stack_api.providers.datatypes import HealthResponse
+from llama_stack_api.datatypes import HealthResponse
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/providers/__init__.py

@@ -1,7 +0,0 @@
-# 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 .datatypes import *  # noqa: F403

src/llama-stack-api/llama_stack_api/rag_tool.py

@@ -9,7 +9,7 @@ from typing import Annotated, Any, Literal
 
 from pydantic import BaseModel, Field, field_validator
 
-from llama_stack_api.apis.common.content_types import URL, InterleavedContent
+from llama_stack_api.common.content_types import URL, InterleavedContent
 
 
 class RRFRanker(BaseModel):

src/llama-stack-api/llama_stack_api/safety.py

@@ -9,11 +9,11 @@ from typing import Any, Protocol, runtime_checkable
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.inference import OpenAIMessageParam
-from llama_stack_api.apis.shields import Shield
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.inference import OpenAIMessageParam
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.shields import Shield
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/scoring.py

@@ -8,9 +8,9 @@ from typing import Any, Protocol, runtime_checkable
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.scoring_functions import ScoringFn, ScoringFnParams
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.scoring_functions import ScoringFn, ScoringFnParams
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 # mapping of metric to value
 ScoringResultRow = dict[str, Any]

src/llama-stack-api/llama_stack_api/scoring_functions.py

@@ -16,10 +16,10 @@ from typing import (
 
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.type_system import ParamType
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.type_system import ParamType
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, register_schema, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 # Perhaps more structure can be imposed on these functions. Maybe they could be associated

src/llama-stack-api/llama_stack_api/shields.py

@@ -8,10 +8,10 @@ from typing import Any, Literal, Protocol, runtime_checkable
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 class CommonShieldFields(BaseModel):

src/llama-stack-api/llama_stack_api/tools.py

@@ -10,11 +10,11 @@ from typing import Any, Literal, Protocol
 from pydantic import BaseModel
 from typing_extensions import runtime_checkable
 
-from llama_stack_api.apis.common.content_types import URL, InterleavedContent
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.resource import Resource, ResourceType
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.content_types import URL, InterleavedContent
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.resource import Resource, ResourceType
 from llama_stack_api.schema_utils import json_schema_type, webmethod
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/vector_io.py

@@ -13,12 +13,12 @@ from typing import Annotated, Any, Literal, Protocol, runtime_checkable
 from fastapi import Body, Query
 from pydantic import BaseModel, Field
 
-from llama_stack_api.apis.common.tracing import telemetry_traceable
-from llama_stack_api.apis.inference import InterleavedContent
-from llama_stack_api.apis.vector_stores import VectorStore
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1
+from llama_stack_api.common.tracing import telemetry_traceable
+from llama_stack_api.inference import InterleavedContent
 from llama_stack_api.schema_utils import json_schema_type, webmethod
 from llama_stack_api.strong_typing.schema import register_schema
+from llama_stack_api.vector_stores import VectorStore
+from llama_stack_api.version import LLAMA_STACK_API_V1
 
 
 @json_schema_type

src/llama-stack-api/llama_stack_api/vector_stores.py

@@ -8,7 +8,7 @@ from typing import Literal
 
 from pydantic import BaseModel
 
-from llama_stack_api.apis.resource import Resource, ResourceType
+from llama_stack_api.resource import Resource, ResourceType
 
 
 # Internal resource type for storing the vector store routing and other information

src/llama_stack/cli/stack/_list_deps.py

@@ -9,7 +9,7 @@ import sys
 from pathlib import Path
 
 import yaml
-from llama_stack_api.providers.datatypes import Api
+from llama_stack_api.datatypes import Api
 from termcolor import cprint
 
 from llama_stack.cli.stack.utils import ImageType

src/llama_stack/cli/stack/utils.py

@@ -11,7 +11,7 @@ from functools import lru_cache
 from pathlib import Path
 
 import yaml
-from llama_stack_api.providers.datatypes import Api
+from llama_stack_api.datatypes import Api
 from termcolor import cprint
 
 from llama_stack.core.datatypes import (

src/llama_stack/core/build.py

@@ -6,7 +6,7 @@
 
 import sys
 
-from llama_stack_api.providers.datatypes import Api
+from llama_stack_api.datatypes import Api
 from pydantic import BaseModel
 from termcolor import cprint
 

src/llama_stack/core/client.py

@@ -12,7 +12,7 @@ from enum import Enum
 from typing import Any, Union, get_args, get_origin
 
 import httpx
-from llama_stack_api.providers.datatypes import RemoteProviderConfig
+from llama_stack_api.datatypes import RemoteProviderConfig
 from pydantic import BaseModel, parse_obj_as
 from termcolor import cprint
 

src/llama_stack/core/configure.py

@@ -6,7 +6,7 @@
 import textwrap
 from typing import Any
 
-from llama_stack_api.providers.datatypes import Api, ProviderSpec
+from llama_stack_api.datatypes import Api, ProviderSpec
 
 from llama_stack.core.datatypes import (
     LLAMA_STACK_RUN_CONFIG_VERSION,

src/llama_stack/core/conversations/conversations.py

@@ -8,7 +8,7 @@ import secrets
 import time
 from typing import Any, Literal
 
-from llama_stack_api.apis.conversations.conversations import (
+from llama_stack_api.conversations import (
     Conversation,
     ConversationDeletedResource,
     ConversationItem,

src/llama_stack/core/datatypes.py

@@ -9,21 +9,21 @@ from pathlib import Path
 from typing import Annotated, Any, Literal, Self
 from urllib.parse import urlparse
 
-from llama_stack_api.apis.benchmarks import Benchmark, BenchmarkInput
-from llama_stack_api.apis.datasetio import DatasetIO
-from llama_stack_api.apis.datasets import Dataset, DatasetInput
-from llama_stack_api.apis.eval import Eval
-from llama_stack_api.apis.inference import Inference
-from llama_stack_api.apis.models import Model, ModelInput
-from llama_stack_api.apis.resource import Resource
-from llama_stack_api.apis.safety import Safety
-from llama_stack_api.apis.scoring import Scoring
-from llama_stack_api.apis.scoring_functions import ScoringFn, ScoringFnInput
-from llama_stack_api.apis.shields import Shield, ShieldInput
-from llama_stack_api.apis.tools import ToolGroup, ToolGroupInput, ToolRuntime
-from llama_stack_api.apis.vector_io import VectorIO
-from llama_stack_api.apis.vector_stores import VectorStore, VectorStoreInput
-from llama_stack_api.providers.datatypes import Api, ProviderSpec
+from llama_stack_api.benchmarks import Benchmark, BenchmarkInput
+from llama_stack_api.datasetio import DatasetIO
+from llama_stack_api.datasets import Dataset, DatasetInput
+from llama_stack_api.datatypes import Api, ProviderSpec
+from llama_stack_api.eval import Eval
+from llama_stack_api.inference import Inference
+from llama_stack_api.models import Model, ModelInput
+from llama_stack_api.resource import Resource
+from llama_stack_api.safety import Safety
+from llama_stack_api.scoring import Scoring
+from llama_stack_api.scoring_functions import ScoringFn, ScoringFnInput
+from llama_stack_api.shields import Shield, ShieldInput
+from llama_stack_api.tools import ToolGroup, ToolGroupInput, ToolRuntime
+from llama_stack_api.vector_io import VectorIO
+from llama_stack_api.vector_stores import VectorStore, VectorStoreInput
 from pydantic import BaseModel, Field, field_validator, model_validator
 
 from llama_stack.core.access_control.datatypes import AccessRule

src/llama_stack/core/distribution.py

@@ -10,7 +10,7 @@ import os
 from typing import Any
 
 import yaml
-from llama_stack_api.providers.datatypes import (
+from llama_stack_api.datatypes import (
     Api,
     InlineProviderSpec,
     ProviderSpec,

src/llama_stack/core/external.py

@@ -6,7 +6,7 @@
 
 
 import yaml
-from llama_stack_api.apis.datatypes import Api, ExternalApiSpec
+from llama_stack_api.datatypes import Api, ExternalApiSpec
 
 from llama_stack.core.datatypes import BuildConfig, StackRunConfig
 from llama_stack.log import get_logger

src/llama_stack/core/inspect.py

@@ -6,14 +6,14 @@
 
 from importlib.metadata import version
 
-from llama_stack_api.apis.inspect import (
+from llama_stack_api.datatypes import HealthStatus
+from llama_stack_api.inspect import (
     HealthInfo,
     Inspect,
     ListRoutesResponse,
     RouteInfo,
     VersionInfo,
 )
-from llama_stack_api.providers.datatypes import HealthStatus
 from pydantic import BaseModel
 
 from llama_stack.core.datatypes import StackRunConfig

src/llama_stack/core/prompts/prompts.py

@@ -7,7 +7,7 @@
 import json
 from typing import Any
 
-from llama_stack_api.apis.prompts import ListPromptsResponse, Prompt, Prompts
+from llama_stack_api.prompts import ListPromptsResponse, Prompt, Prompts
 from pydantic import BaseModel
 
 from llama_stack.core.datatypes import StackRunConfig

src/llama_stack/core/providers.py

@@ -7,8 +7,8 @@
 import asyncio
 from typing import Any
 
-from llama_stack_api.apis.providers import ListProvidersResponse, ProviderInfo, Providers
-from llama_stack_api.providers.datatypes import HealthResponse, HealthStatus
+from llama_stack_api.datatypes import HealthResponse, HealthStatus
+from llama_stack_api.providers import ListProvidersResponse, ProviderInfo, Providers
 from pydantic import BaseModel
 
 from llama_stack.log import get_logger

src/llama_stack/core/resolver.py

@@ -8,33 +8,17 @@ import importlib.metadata
 import inspect
 from typing import Any
 
-from llama_stack_api.apis.agents import Agents
-from llama_stack_api.apis.batches import Batches
-from llama_stack_api.apis.benchmarks import Benchmarks
-from llama_stack_api.apis.conversations import Conversations
-from llama_stack_api.apis.datasetio import DatasetIO
-from llama_stack_api.apis.datasets import Datasets
-from llama_stack_api.apis.datatypes import ExternalApiSpec
-from llama_stack_api.apis.eval import Eval
-from llama_stack_api.apis.files import Files
-from llama_stack_api.apis.inference import Inference, InferenceProvider
-from llama_stack_api.apis.inspect import Inspect
-from llama_stack_api.apis.models import Models
-from llama_stack_api.apis.post_training import PostTraining
-from llama_stack_api.apis.prompts import Prompts
-from llama_stack_api.apis.providers import Providers as ProvidersAPI
-from llama_stack_api.apis.safety import Safety
-from llama_stack_api.apis.scoring import Scoring
-from llama_stack_api.apis.scoring_functions import ScoringFunctions
-from llama_stack_api.apis.shields import Shields
-from llama_stack_api.apis.tools import ToolGroups, ToolRuntime
-from llama_stack_api.apis.vector_io import VectorIO
-from llama_stack_api.apis.vector_stores import VectorStore
-from llama_stack_api.apis.version import LLAMA_STACK_API_V1ALPHA
-from llama_stack_api.providers.datatypes import (
+from llama_stack_api.agents import Agents
+from llama_stack_api.batches import Batches
+from llama_stack_api.benchmarks import Benchmarks
+from llama_stack_api.conversations import Conversations
+from llama_stack_api.datasetio import DatasetIO
+from llama_stack_api.datasets import Datasets
+from llama_stack_api.datatypes import (
     Api,
     BenchmarksProtocolPrivate,
     DatasetsProtocolPrivate,
+    ExternalApiSpec,
     ModelsProtocolPrivate,
     ProviderSpec,
     RemoteProviderConfig,
@@ -43,6 +27,22 @@ from llama_stack_api.providers.datatypes import (
     ShieldsProtocolPrivate,
     ToolGroupsProtocolPrivate,
 )
+from llama_stack_api.eval import Eval
+from llama_stack_api.files import Files
+from llama_stack_api.inference import Inference, InferenceProvider
+from llama_stack_api.inspect import Inspect
+from llama_stack_api.models import Models
+from llama_stack_api.post_training import PostTraining
+from llama_stack_api.prompts import Prompts
+from llama_stack_api.providers import Providers as ProvidersAPI
+from llama_stack_api.safety import Safety
+from llama_stack_api.scoring import Scoring
+from llama_stack_api.scoring_functions import ScoringFunctions
+from llama_stack_api.shields import Shields
+from llama_stack_api.tools import ToolGroups, ToolRuntime
+from llama_stack_api.vector_io import VectorIO
+from llama_stack_api.vector_stores import VectorStore
+from llama_stack_api.version import LLAMA_STACK_API_V1ALPHA
 
 from llama_stack.core.client import get_client_impl
 from llama_stack.core.datatypes import (

src/llama_stack/core/routers/__init__.py

@@ -6,7 +6,7 @@
 
 from typing import Any
 
-from llama_stack_api.providers.datatypes import Api, RoutingTable
+from llama_stack_api.datatypes import Api, RoutingTable
 
 from llama_stack.core.datatypes import (
     AccessRule,

src/llama_stack/core/routers/datasets.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
-from llama_stack_api.apis.common.responses import PaginatedResponse
-from llama_stack_api.apis.datasetio import DatasetIO
-from llama_stack_api.apis.datasets import DatasetPurpose, DataSource
-from llama_stack_api.providers.datatypes import RoutingTable
+from llama_stack_api.common.responses import PaginatedResponse
+from llama_stack_api.datasetio import DatasetIO
+from llama_stack_api.datasets import DatasetPurpose, DataSource
+from llama_stack_api.datatypes import RoutingTable
 
 from llama_stack.log import get_logger
 

src/llama_stack/core/routers/eval_scoring.py

@@ -6,14 +6,14 @@
 
 from typing import Any
 
-from llama_stack_api.apis.eval import BenchmarkConfig, Eval, EvaluateResponse, Job
-from llama_stack_api.apis.scoring import (
+from llama_stack_api.datatypes import RoutingTable
+from llama_stack_api.eval import BenchmarkConfig, Eval, EvaluateResponse, Job
+from llama_stack_api.scoring import (
     ScoreBatchResponse,
     ScoreResponse,
     Scoring,
     ScoringFnParams,
 )
-from llama_stack_api.providers.datatypes import RoutingTable
 
 from llama_stack.log import get_logger
 

src/llama_stack/core/routers/inference.py

@@ -11,13 +11,16 @@ from datetime import UTC, datetime
 from typing import Annotated, Any
 
 from fastapi import Body
-from llama_stack_api.apis.common.errors import ModelNotFoundError, ModelTypeError
-from llama_stack_api.apis.inference import (
+from llama_stack_api.common.errors import ModelNotFoundError, ModelTypeError
+from llama_stack_api.datatypes import HealthResponse, HealthStatus, RoutingTable
+from llama_stack_api.inference import (
     Inference,
     ListOpenAIChatCompletionResponse,
     OpenAIAssistantMessageParam,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
+    OpenAIChatCompletionContentPartImageParam,
+    OpenAIChatCompletionContentPartTextParam,
     OpenAIChatCompletionRequestWithExtraBody,
     OpenAIChatCompletionToolCall,
     OpenAIChatCompletionToolCallFunction,
@@ -32,12 +35,7 @@ from llama_stack_api.apis.inference import (
     Order,
     RerankResponse,
 )
-from llama_stack_api.apis.inference.inference import (
-    OpenAIChatCompletionContentPartImageParam,
-    OpenAIChatCompletionContentPartTextParam,
-)
-from llama_stack_api.apis.models import ModelType
-from llama_stack_api.providers.datatypes import HealthResponse, HealthStatus, RoutingTable
+from llama_stack_api.models import ModelType
 from openai.types.chat import ChatCompletionToolChoiceOptionParam as OpenAIChatCompletionToolChoiceOptionParam
 from openai.types.chat import ChatCompletionToolParam as OpenAIChatCompletionToolParam
 from pydantic import TypeAdapter

src/llama_stack/core/routers/safety.py

@@ -6,11 +6,10 @@
 
 from typing import Any
 
-from llama_stack_api.apis.inference import OpenAIMessageParam
-from llama_stack_api.apis.safety import RunShieldResponse, Safety
-from llama_stack_api.apis.safety.safety import ModerationObject
-from llama_stack_api.apis.shields import Shield
-from llama_stack_api.providers.datatypes import RoutingTable
+from llama_stack_api.datatypes import RoutingTable
+from llama_stack_api.inference import OpenAIMessageParam
+from llama_stack_api.safety import ModerationObject, RunShieldResponse, Safety
+from llama_stack_api.shields import Shield
 
 from llama_stack.core.datatypes import SafetyConfig
 from llama_stack.log import get_logger

src/llama_stack/core/routers/tool_runtime.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
-from llama_stack_api.apis.common.content_types import (
+from llama_stack_api.common.content_types import (
     URL,
 )
-from llama_stack_api.apis.tools import (
+from llama_stack_api.tools import (
     ListToolDefsResponse,
     ToolRuntime,
 )

src/llama_stack/core/routers/vector_io.py

@@ -9,9 +9,10 @@ import uuid
 from typing import Annotated, Any
 
 from fastapi import Body
-from llama_stack_api.apis.common.content_types import InterleavedContent
-from llama_stack_api.apis.models import ModelType
-from llama_stack_api.apis.vector_io import (
+from llama_stack_api.common.content_types import InterleavedContent
+from llama_stack_api.datatypes import HealthResponse, HealthStatus, RoutingTable
+from llama_stack_api.models import ModelType
+from llama_stack_api.vector_io import (
     Chunk,
     OpenAICreateVectorStoreFileBatchRequestWithExtraBody,
     OpenAICreateVectorStoreRequestWithExtraBody,
@@ -32,7 +33,6 @@ from llama_stack_api.apis.vector_io import (
     VectorStoreObject,
     VectorStoreSearchResponsePage,
 )
-from llama_stack_api.providers.datatypes import HealthResponse, HealthStatus, RoutingTable
 
 from llama_stack.core.datatypes import VectorStoresConfig
 from llama_stack.log import get_logger

src/llama_stack/core/routing_tables/benchmarks.py

@@ -6,7 +6,7 @@
 
 from typing import Any
 
-from llama_stack_api.apis.benchmarks import Benchmark, Benchmarks, ListBenchmarksResponse
+from llama_stack_api.benchmarks import Benchmark, Benchmarks, ListBenchmarksResponse
 
 from llama_stack.core.datatypes import (
     BenchmarkWithOwner,

src/llama_stack/core/routing_tables/common.py

@@ -6,10 +6,10 @@
 
 from typing import Any
 
-from llama_stack_api.apis.common.errors import ModelNotFoundError
-from llama_stack_api.apis.models import Model
-from llama_stack_api.apis.resource import ResourceType
-from llama_stack_api.providers.datatypes import Api, RoutingTable
+from llama_stack_api.common.errors import ModelNotFoundError
+from llama_stack_api.datatypes import Api, RoutingTable
+from llama_stack_api.models import Model
+from llama_stack_api.resource import ResourceType
 
 from llama_stack.core.access_control.access_control import AccessDeniedError, is_action_allowed
 from llama_stack.core.access_control.datatypes import Action

src/llama_stack/core/routing_tables/datasets.py

@@ -7,8 +7,8 @@
 import uuid
 from typing import Any
 
-from llama_stack_api.apis.common.errors import DatasetNotFoundError
-from llama_stack_api.apis.datasets import (
+from llama_stack_api.common.errors import DatasetNotFoundError
+from llama_stack_api.datasets import (
     Dataset,
     DatasetPurpose,
     Datasets,
@@ -18,7 +18,7 @@ from llama_stack_api.apis.datasets import (
     RowsDataSource,
     URIDataSource,
 )
-from llama_stack_api.apis.resource import ResourceType
+from llama_stack_api.resource import ResourceType
 
 from llama_stack.core.datatypes import (
     DatasetWithOwner,

src/llama_stack/core/routing_tables/models.py

@@ -7,8 +7,8 @@
 import time
 from typing import Any
 
-from llama_stack_api.apis.common.errors import ModelNotFoundError
-from llama_stack_api.apis.models import (
+from llama_stack_api.common.errors import ModelNotFoundError
+from llama_stack_api.models import (
     ListModelsResponse,
     Model,
     Models,

src/llama_stack/core/routing_tables/scoring_functions.py

@@ -4,9 +4,9 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack_api.apis.common.type_system import ParamType
-from llama_stack_api.apis.resource import ResourceType
-from llama_stack_api.apis.scoring_functions import (
+from llama_stack_api.common.type_system import ParamType
+from llama_stack_api.resource import ResourceType
+from llama_stack_api.scoring_functions import (
     ListScoringFunctionsResponse,
     ScoringFn,
     ScoringFnParams,

src/llama_stack/core/routing_tables/shields.py

@@ -6,8 +6,8 @@
 
 from typing import Any
 
-from llama_stack_api.apis.resource import ResourceType
-from llama_stack_api.apis.shields import ListShieldsResponse, Shield, Shields
+from llama_stack_api.resource import ResourceType
+from llama_stack_api.shields import ListShieldsResponse, Shield, Shields
 
 from llama_stack.core.datatypes import (
     ShieldWithOwner,

src/llama_stack/core/routing_tables/toolgroups.py

@@ -6,9 +6,9 @@
 
 from typing import Any
 
-from llama_stack_api.apis.common.content_types import URL
-from llama_stack_api.apis.common.errors import ToolGroupNotFoundError
-from llama_stack_api.apis.tools import ListToolDefsResponse, ListToolGroupsResponse, ToolDef, ToolGroup, ToolGroups
+from llama_stack_api.common.content_types import URL
+from llama_stack_api.common.errors import ToolGroupNotFoundError
+from llama_stack_api.tools import ListToolDefsResponse, ListToolGroupsResponse, ToolDef, ToolGroup, ToolGroups
 
 from llama_stack.core.datatypes import AuthenticationRequiredError, ToolGroupWithOwner
 from llama_stack.log import get_logger

src/llama_stack/core/routing_tables/vector_stores.py

@@ -6,12 +6,12 @@
 
 from typing import Any
 
-from llama_stack_api.apis.common.errors import ModelNotFoundError, ModelTypeError
-from llama_stack_api.apis.models import ModelType
-from llama_stack_api.apis.resource import ResourceType
+from llama_stack_api.common.errors import ModelNotFoundError, ModelTypeError
+from llama_stack_api.models import ModelType
+from llama_stack_api.resource import ResourceType
 
 # Removed VectorStores import to avoid exposing public API
-from llama_stack_api.apis.vector_io.vector_io import (
+from llama_stack_api.vector_io import (
     SearchRankingOptions,
     VectorStoreChunkingStrategy,
     VectorStoreDeleteResponse,

src/llama_stack/core/server/auth_providers.py

@@ -11,7 +11,7 @@ from urllib.parse import parse_qs, urljoin, urlparse
 
 import httpx
 import jwt
-from llama_stack_api.apis.common.errors import TokenValidationError
+from llama_stack_api.common.errors import TokenValidationError
 from pydantic import BaseModel, Field
 
 from llama_stack.core.datatypes import (

src/llama_stack/core/server/routes.py

@@ -10,7 +10,7 @@ from collections.abc import Callable
 from typing import Any
 
 from aiohttp import hdrs
-from llama_stack_api.apis.datatypes import Api, ExternalApiSpec
+from llama_stack_api.datatypes import Api, ExternalApiSpec
 from llama_stack_api.schema_utils import WebMethod
 from starlette.routing import Route
 

src/llama_stack/core/server/server.py

@@ -28,9 +28,9 @@ from fastapi import Path as FastapiPath
 from fastapi.exceptions import RequestValidationError
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import JSONResponse, StreamingResponse
-from llama_stack_api.apis.common.errors import ConflictError, ResourceNotFoundError
-from llama_stack_api.apis.common.responses import PaginatedResponse
-from llama_stack_api.providers.datatypes import Api
+from llama_stack_api.common.errors import ConflictError, ResourceNotFoundError
+from llama_stack_api.common.responses import PaginatedResponse
+from llama_stack_api.datatypes import Api
 from openai import BadRequestError
 from pydantic import BaseModel, ValidationError
 

src/llama_stack/core/stack.py

@@ -12,27 +12,27 @@ import tempfile
 from typing import Any
 
 import yaml
-from llama_stack_api.apis.agents import Agents
-from llama_stack_api.apis.batches import Batches
-from llama_stack_api.apis.benchmarks import Benchmarks
-from llama_stack_api.apis.conversations import Conversations
-from llama_stack_api.apis.datasetio import DatasetIO
-from llama_stack_api.apis.datasets import Datasets
-from llama_stack_api.apis.eval import Eval
-from llama_stack_api.apis.files import Files
-from llama_stack_api.apis.inference import Inference
-from llama_stack_api.apis.inspect import Inspect
-from llama_stack_api.apis.models import Models
-from llama_stack_api.apis.post_training import PostTraining
-from llama_stack_api.apis.prompts import Prompts
-from llama_stack_api.apis.providers import Providers
-from llama_stack_api.apis.safety import Safety
-from llama_stack_api.apis.scoring import Scoring
-from llama_stack_api.apis.scoring_functions import ScoringFunctions
-from llama_stack_api.apis.shields import Shields
-from llama_stack_api.apis.tools import ToolGroups, ToolRuntime
-from llama_stack_api.apis.vector_io import VectorIO
-from llama_stack_api.providers.datatypes import Api
+from llama_stack_api.agents import Agents
+from llama_stack_api.batches import Batches
+from llama_stack_api.benchmarks import Benchmarks
+from llama_stack_api.conversations import Conversations
+from llama_stack_api.datasetio import DatasetIO
+from llama_stack_api.datasets import Datasets
+from llama_stack_api.datatypes import Api
+from llama_stack_api.eval import Eval
+from llama_stack_api.files import Files
+from llama_stack_api.inference import Inference
+from llama_stack_api.inspect import Inspect
+from llama_stack_api.models import Models
+from llama_stack_api.post_training import PostTraining
+from llama_stack_api.prompts import Prompts
+from llama_stack_api.providers import Providers
+from llama_stack_api.safety import Safety
+from llama_stack_api.scoring import Scoring
+from llama_stack_api.scoring_functions import ScoringFunctions
+from llama_stack_api.shields import Shields
+from llama_stack_api.tools import ToolGroups, ToolRuntime
+from llama_stack_api.vector_io import VectorIO
 
 from llama_stack.core.conversations.conversations import ConversationServiceConfig, ConversationServiceImpl
 from llama_stack.core.datatypes import Provider, SafetyConfig, StackRunConfig, VectorStoresConfig

src/llama_stack/distributions/dell/dell.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_api.apis.models import ModelType
+from llama_stack_api.models import ModelType
 
 from llama_stack.core.datatypes import (
     BuildProvider,

src/llama_stack/distributions/meta-reference-gpu/meta_reference.py

@@ -6,7 +6,7 @@
 
 from pathlib import Path
 
-from llama_stack_api.apis.models import ModelType
+from llama_stack_api.models import ModelType
 
 from llama_stack.core.datatypes import (
     BuildProvider,