add tools api with a stub provider impl

llama_stack/apis/resource.py

@@ -18,6 +18,7 @@ class ResourceType(Enum):
     dataset = "dataset"
     scoring_function = "scoring_function"
     eval_task = "eval_task"
+    tool = "tool"
 
 
 class Resource(BaseModel):

llama_stack/apis/tools/tools.py

@@ -7,12 +7,11 @@
 from typing import Any, Dict, List, Literal, Optional
 
 from llama_models.llama3.api.datatypes import ToolPromptFormat
-
+from llama_models.schema_utils import json_schema_type, webmethod
-from llama_models.schema_utils import json_schema_type
 from pydantic import BaseModel, Field
 from typing_extensions import Protocol, runtime_checkable
 
-from llama_stack.apis.resource import Resource
+from llama_stack.apis.resource import Resource, ResourceType
 from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
 
 
@@ -39,7 +38,7 @@ class ToolReturn(BaseModel):
 class Tool(Resource):
     """Represents a tool that can be provided by different providers"""
 
-    resource_type: Literal["tool"] = "tool"
+    type: Literal[ResourceType.tool.value] = ResourceType.tool.value
     name: str
     description: str
     parameters: List[ToolParameter]
@@ -53,6 +52,7 @@ class Tool(Resource):
 @runtime_checkable
 @trace_protocol
 class Tools(Protocol):
+    @webmethod(route="/tools/register", method="POST")
     async def register_tool(
         self,
         tool_id: str,
@@ -60,27 +60,33 @@ class Tools(Protocol):
         description: str,
         parameters: List[ToolParameter],
         returns: ToolReturn,
+        provider_id: Optional[str] = None,
         provider_metadata: Optional[Dict[str, Any]] = None,
         tool_prompt_format: Optional[ToolPromptFormat] = None,
     ) -> Tool:
         """Register a tool with provider-specific metadata"""
         ...
 
+    @webmethod(route="/tools/get", method="GET")
     async def get_tool(
         self,
-        identifier: str,
+        tool_id: str,
     ) -> Tool: ...
 
-    async def list_tools(
+    @webmethod(route="/tools/list", method="GET")
-        self,
+    async def list_tools(self) -> List[Tool]:
-        provider_id: Optional[str] = None,
-    ) -> List[Tool]:
         """List tools with optional provider"""
 
+    @webmethod(route="/tools/unregister", method="POST")
+    async def unregister_tool(self, tool_id: str) -> None:
+        """Unregister a tool"""
+        ...
+
 
 @runtime_checkable
 @trace_protocol
 class ToolRuntime(Protocol):
-    def invoke_tool(self, tool_id: str, args: Dict[str, Any]) -> Any:
+    @webmethod(route="/tool-runtime/invoke", method="POST")
+    async def invoke_tool(self, tool_id: str, args: Dict[str, Any]) -> Any:
         """Run a tool with the given arguments"""
         ...

llama_stack/distribution/distribution.py

@@ -47,6 +47,10 @@ def builtin_automatically_routed_apis() -> List[AutoRoutedApiInfo]:
             routing_table_api=Api.eval_tasks,
             router_api=Api.eval,
         ),
+        AutoRoutedApiInfo(
+            routing_table_api=Api.tools,
+            router_api=Api.tool_runtime,
+        ),
     ]
 
 

llama_stack/distribution/resolver.py

@@ -30,7 +30,7 @@ from llama_stack.apis.scoring import Scoring
 from llama_stack.apis.scoring_functions import ScoringFunctions
 from llama_stack.apis.shields import Shields
 from llama_stack.apis.telemetry import Telemetry
-from llama_stack.apis.tools import Tools
+from llama_stack.apis.tools import ToolRuntime, Tools
 from llama_stack.distribution.client import get_client_impl
 from llama_stack.distribution.distribution import builtin_automatically_routed_apis
 from llama_stack.distribution.store import DistributionRegistry
@@ -61,6 +61,8 @@ def api_protocol_map() -> Dict[Api, Any]:
         Api.eval: Eval,
         Api.eval_tasks: EvalTasks,
         Api.post_training: PostTraining,
+        Api.tools: Tools,
+        Api.tool_runtime: ToolRuntime,
     }
 
 

llama_stack/distribution/routers/__init__.py

@@ -17,6 +17,7 @@ from .routing_tables import (
     ModelsRoutingTable,
     ScoringFunctionsRoutingTable,
     ShieldsRoutingTable,
+    ToolsRoutingTable,
 )
 
 
@@ -33,6 +34,7 @@ async def get_routing_table_impl(
         "datasets": DatasetsRoutingTable,
         "scoring_functions": ScoringFunctionsRoutingTable,
         "eval_tasks": EvalTasksRoutingTable,
+        "tools": ToolsRoutingTable,
     }
 
     if api.value not in api_to_tables:
@@ -51,6 +53,7 @@ async def get_auto_router_impl(api: Api, routing_table: RoutingTable, _deps) ->
         MemoryRouter,
         SafetyRouter,
         ScoringRouter,
+        ToolRuntimeRouter,
     )
 
     api_to_routers = {
@@ -60,6 +63,7 @@ async def get_auto_router_impl(api: Api, routing_table: RoutingTable, _deps) ->
         "datasetio": DatasetIORouter,
         "scoring": ScoringRouter,
         "eval": EvalRouter,
+        "tool_runtime": ToolRuntimeRouter,
     }
     if api.value not in api_to_routers:
         raise ValueError(f"API {api.value} not found in router map")

llama_stack/distribution/routers/routers.py

@@ -15,6 +15,7 @@ from llama_stack.apis.safety import *  # noqa: F403
 from llama_stack.apis.datasetio import *  # noqa: F403
 from llama_stack.apis.scoring import *  # noqa: F403
 from llama_stack.apis.eval import *  # noqa: F403
+from llama_stack.apis.tools import *  # noqa: F403
 
 
 class MemoryRouter(Memory):
@@ -372,3 +373,23 @@ class EvalRouter(Eval):
             task_id,
             job_id,
         )
+
+
+class ToolRuntimeRouter(ToolRuntime):
+    def __init__(
+        self,
+        routing_table: RoutingTable,
+    ) -> None:
+        self.routing_table = routing_table
+
+    async def initialize(self) -> None:
+        pass
+
+    async def shutdown(self) -> None:
+        pass
+
+    async def invoke_tool(self, tool_id: str, args: Dict[str, Any]) -> Any:
+        return await self.routing_table.get_provider_impl(tool_id).invoke_tool(
+            tool_id=tool_id,
+            args=args,
+        )

llama_stack/distribution/routers/routing_tables.py

@@ -15,7 +15,7 @@ from llama_stack.apis.shields import *  # noqa: F403
 from llama_stack.apis.memory_banks import *  # noqa: F403
 from llama_stack.apis.datasets import *  # noqa: F403
 from llama_stack.apis.eval_tasks import *  # noqa: F403
-
+from llama_stack.apis.tools import *  # noqa: F403
 from llama_stack.apis.common.content_types import URL
 
 from llama_stack.apis.common.type_system import ParamType
@@ -45,6 +45,8 @@ async def register_object_with_provider(obj: RoutableObject, p: Any) -> Routable
         return await p.register_scoring_function(obj)
     elif api == Api.eval:
         return await p.register_eval_task(obj)
+    elif api == Api.tool_runtime:
+        return await p.register_tool(obj)
     else:
         raise ValueError(f"Unknown API {api} for registering object with provider")
 
@@ -57,6 +59,8 @@ async def unregister_object_from_provider(obj: RoutableObject, p: Any) -> None:
         return await p.unregister_model(obj.identifier)
     elif api == Api.datasetio:
         return await p.unregister_dataset(obj.identifier)
+    elif api == Api.tool_runtime:
+        return await p.unregister_tool(obj.identifier)
     else:
         raise ValueError(f"Unregister not supported for {api}")
 
@@ -461,3 +465,51 @@ class EvalTasksRoutingTable(CommonRoutingTableImpl, EvalTasks):
             provider_resource_id=provider_eval_task_id,
         )
         await self.register_object(eval_task)
+
+
+class ToolsRoutingTable(CommonRoutingTableImpl, Tools):
+    async def list_tools(self) -> List[Tool]:
+        return await self.get_all_with_type("tool")
+
+    async def get_tool(self, tool_id: str) -> Tool:
+        return await self.get_object_by_identifier("tool", tool_id)
+
+    async def register_tool(
+        self,
+        tool_id: str,
+        name: str,
+        description: str,
+        parameters: List[ToolParameter],
+        returns: ToolReturn,
+        provider_id: Optional[str] = None,
+        provider_metadata: Optional[Dict[str, Any]] = None,
+        tool_prompt_format: Optional[ToolPromptFormat] = None,
+    ) -> None:
+        if provider_metadata is None:
+            provider_metadata = {}
+        if tool_prompt_format is None:
+            tool_prompt_format = ToolPromptFormat.json
+        if provider_id is None:
+            if len(self.impls_by_provider_id) == 1:
+                provider_id = list(self.impls_by_provider_id.keys())[0]
+            else:
+                raise ValueError(
+                    "No provider specified and multiple providers available. Please specify a provider_id."
+                )
+        tool = Tool(
+            identifier=tool_id,
+            name=name,
+            description=description,
+            parameters=parameters,
+            returns=returns,
+            provider_id=provider_id,
+            provider_metadata=provider_metadata,
+            tool_prompt_format=tool_prompt_format,
+        )
+        await self.register_object(tool)
+
+    async def unregister_tool(self, tool_id: str) -> None:
+        tool = await self.get_tool(tool_id)
+        if tool is None:
+            raise ValueError(f"Tool {tool_id} not found")
+        await self.unregister_object(tool)

llama_stack/providers/datatypes.py

@@ -30,6 +30,7 @@ class Api(Enum):
     scoring = "scoring"
     eval = "eval"
     post_training = "post_training"
+    tool_runtime = "tool_runtime"
 
     telemetry = "telemetry"
 
@@ -39,6 +40,7 @@ class Api(Enum):
     datasets = "datasets"
     scoring_functions = "scoring_functions"
     eval_tasks = "eval_tasks"
+    tools = "tools"
 
     # built-in API
     inspect = "inspect"
@@ -79,6 +81,8 @@ class EvalTasksProtocolPrivate(Protocol):
 class ToolsProtocolPrivate(Protocol):
     async def register_tool(self, tool: Tool) -> None: ...
 
+    async def unregister_tool(self, tool_id: str) -> None: ...
+
 
 @json_schema_type
 class ProviderSpec(BaseModel):

llama_stack/providers/inline/tool_runtime/meta_reference/__init__.py

@@ -0,0 +1,14 @@
+# 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 .config import MetaReferenceToolRuntimeConfig
+from .meta_reference import MetaReferenceToolRuntimeImpl
+
+
+async def get_provider_impl(config: MetaReferenceToolRuntimeConfig, _deps):
+    impl = MetaReferenceToolRuntimeImpl(config)
+    await impl.initialize()
+    return impl

llama_stack/providers/inline/tool_runtime/meta_reference/config.py

@@ -7,5 +7,5 @@
 from pydantic import BaseModel
 
 
-class MetaReferenceToolConfig(BaseModel):
+class MetaReferenceToolRuntimeConfig(BaseModel):
     pass

llama_stack/providers/inline/tool_runtime/meta_reference/meta_reference.py

@@ -0,0 +1,30 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from typing import Any, Dict
+
+from llama_stack.apis.tools import Tool, ToolRuntime
+from llama_stack.providers.datatypes import ToolsProtocolPrivate
+
+from .config import MetaReferenceToolRuntimeConfig
+
+
+class MetaReferenceToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime):
+    def __init__(self, config: MetaReferenceToolRuntimeConfig):
+        self.config = config
+
+    async def initialize(self):
+        pass
+
+    async def register_tool(self, tool: Tool):
+        print(f"registering tool {tool.identifier}")
+        pass
+
+    async def unregister_tool(self, tool_id: str) -> None:
+        pass
+
+    async def invoke_tool(self, tool_id: str, args: Dict[str, Any]) -> Any:
+        pass

llama_stack/providers/inline/tools/meta_reference/__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 .meta_reference import *  # noqa: F401 F403

llama_stack/providers/inline/tools/meta_reference/meta_reference.py

@@ -1,17 +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 llama_stack.apis.tools import Tool, Tools
-
-from .config import MetaReferenceToolConfig
-
-
-class MetaReferenceTool(Tools):
-    def __init__(self, config: MetaReferenceToolConfig):
-        self.config = config
-
-    async def register_tool(self, tool: Tool):
-        pass

llama_stack/providers/registry/tool_runtime.py

@@ -0,0 +1,21 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from typing import List
+
+from llama_stack.distribution.datatypes import Api, InlineProviderSpec, ProviderSpec
+
+
+def available_providers() -> List[ProviderSpec]:
+    return [
+        InlineProviderSpec(
+            api=Api.tool_runtime,
+            provider_type="inline::meta-reference",
+            pip_packages=[],
+            module="llama_stack.providers.inline.tool_runtime.meta_reference",
+            config_class="llama_stack.providers.inline.tool_runtime.meta_reference.MetaReferenceToolRuntimeConfig",
+        ),
+    ]