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/core/routers/__init__.py

@@ -6,13 +6,14 @@
 
 from typing import Any
 
+from llama_stack_api import Api, RoutingTable
+
 from llama_stack.core.datatypes import (
     AccessRule,
     RoutedProtocol,
 )
 from llama_stack.core.stack import StackRunConfig
 from llama_stack.core.store import DistributionRegistry
-from llama_stack.providers.datatypes import Api, RoutingTable
 from llama_stack.providers.utils.inference.inference_store import InferenceStore
 
 

src/llama_stack/core/routers/datasets.py

@@ -6,11 +6,9 @@
 
 from typing import Any
 
-from llama_stack.apis.common.responses import PaginatedResponse
-from llama_stack.apis.datasetio import DatasetIO
-from llama_stack.apis.datasets import DatasetPurpose, DataSource
+from llama_stack_api import DatasetIO, DatasetPurpose, DataSource, PaginatedResponse, RoutingTable
+
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import RoutingTable
 
 logger = get_logger(name=__name__, category="core::routers")
 

src/llama_stack/core/routers/eval_scoring.py

@@ -6,15 +6,19 @@
 
 from typing import Any
 
-from llama_stack.apis.eval import BenchmarkConfig, Eval, EvaluateResponse, Job
-from llama_stack.apis.scoring import (
+from llama_stack_api import (
+    BenchmarkConfig,
+    Eval,
+    EvaluateResponse,
+    Job,
+    RoutingTable,
     ScoreBatchResponse,
     ScoreResponse,
     Scoring,
     ScoringFnParams,
 )
+
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import RoutingTable
 
 logger = get_logger(name=__name__, category="core::routers")
 

src/llama_stack/core/routers/inference.py

@@ -11,17 +11,19 @@ from datetime import UTC, datetime
 from typing import Annotated, Any
 
 from fastapi import Body
-from openai.types.chat import ChatCompletionToolChoiceOptionParam as OpenAIChatCompletionToolChoiceOptionParam
-from openai.types.chat import ChatCompletionToolParam as OpenAIChatCompletionToolParam
-from pydantic import TypeAdapter
-
-from llama_stack.apis.common.errors import ModelNotFoundError, ModelTypeError
-from llama_stack.apis.inference import (
+from llama_stack_api import (
+    HealthResponse,
+    HealthStatus,
     Inference,
     ListOpenAIChatCompletionResponse,
+    ModelNotFoundError,
+    ModelType,
+    ModelTypeError,
     OpenAIAssistantMessageParam,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
+    OpenAIChatCompletionContentPartImageParam,
+    OpenAIChatCompletionContentPartTextParam,
     OpenAIChatCompletionRequestWithExtraBody,
     OpenAIChatCompletionToolCall,
     OpenAIChatCompletionToolCallFunction,
@@ -35,18 +37,17 @@ from llama_stack.apis.inference import (
     OpenAIMessageParam,
     Order,
     RerankResponse,
+    RoutingTable,
 )
-from llama_stack.apis.inference.inference import (
-    OpenAIChatCompletionContentPartImageParam,
-    OpenAIChatCompletionContentPartTextParam,
-)
-from llama_stack.apis.models import ModelType
+from openai.types.chat import ChatCompletionToolChoiceOptionParam as OpenAIChatCompletionToolChoiceOptionParam
+from openai.types.chat import ChatCompletionToolParam as OpenAIChatCompletionToolParam
+from pydantic import TypeAdapter
+
 from llama_stack.core.telemetry.telemetry import MetricEvent
 from llama_stack.core.telemetry.tracing import enqueue_event, get_current_span
 from llama_stack.log import get_logger
 from llama_stack.models.llama.llama3.chat_format import ChatFormat
 from llama_stack.models.llama.llama3.tokenizer import Tokenizer
-from llama_stack.providers.datatypes import HealthResponse, HealthStatus, RoutingTable
 from llama_stack.providers.utils.inference.inference_store import InferenceStore
 
 logger = get_logger(name=__name__, category="core::routers")

src/llama_stack/core/routers/safety.py

@@ -6,13 +6,10 @@
 
 from typing import Any
 
-from llama_stack.apis.inference import OpenAIMessageParam
-from llama_stack.apis.safety import RunShieldResponse, Safety
-from llama_stack.apis.safety.safety import ModerationObject
-from llama_stack.apis.shields import Shield
+from llama_stack_api import ModerationObject, OpenAIMessageParam, RoutingTable, RunShieldResponse, Safety, Shield
+
 from llama_stack.core.datatypes import SafetyConfig
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import RoutingTable
 
 logger = get_logger(name=__name__, category="core::routers")
 

src/llama_stack/core/routers/tool_runtime.py

@@ -6,13 +6,12 @@
 
 from typing import Any
 
-from llama_stack.apis.common.content_types import (
+from llama_stack_api import (
     URL,
-)
-from llama_stack.apis.tools import (
     ListToolDefsResponse,
     ToolRuntime,
 )
+
 from llama_stack.log import get_logger
 
 from ..routing_tables.toolgroups import ToolGroupsRoutingTable

src/llama_stack/core/routers/vector_io.py

@@ -9,14 +9,16 @@ import uuid
 from typing import Annotated, Any
 
 from fastapi import Body
-
-from llama_stack.apis.common.content_types import InterleavedContent
-from llama_stack.apis.models import ModelType
-from llama_stack.apis.vector_io import (
+from llama_stack_api import (
     Chunk,
+    HealthResponse,
+    HealthStatus,
+    InterleavedContent,
+    ModelType,
     OpenAICreateVectorStoreFileBatchRequestWithExtraBody,
     OpenAICreateVectorStoreRequestWithExtraBody,
     QueryChunksResponse,
+    RoutingTable,
     SearchRankingOptions,
     VectorIO,
     VectorStoreChunkingStrategy,
@@ -33,9 +35,9 @@ from llama_stack.apis.vector_io import (
     VectorStoreObject,
     VectorStoreSearchResponsePage,
 )
+
 from llama_stack.core.datatypes import VectorStoresConfig
 from llama_stack.log import get_logger
-from llama_stack.providers.datatypes import HealthResponse, HealthStatus, RoutingTable
 
 logger = get_logger(name=__name__, category="core::routers")