aiqtoolkit 1.1.0a20250429__py3-none-any.whl

aiq/builder/workflow_builder.py

@@ -0,0 +1,733 @@
+# SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import dataclasses
+import inspect
+import logging
+import warnings
+from contextlib import AbstractAsyncContextManager
+from contextlib import AsyncExitStack
+from contextlib import asynccontextmanager
+
+from opentelemetry import trace
+from opentelemetry.sdk.trace import TracerProvider
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from opentelemetry.sdk.trace.export import SpanExporter
+
+from aiq.builder.builder import Builder
+from aiq.builder.builder import UserManagerHolder
+from aiq.builder.component_utils import build_dependency_sequence
+from aiq.builder.context import AIQContext
+from aiq.builder.context import AIQContextState
+from aiq.builder.embedder import EmbedderProviderInfo
+from aiq.builder.framework_enum import LLMFrameworkEnum
+from aiq.builder.function import Function
+from aiq.builder.function import LambdaFunction
+from aiq.builder.function_info import FunctionInfo
+from aiq.builder.llm import LLMProviderInfo
+from aiq.builder.retriever import RetrieverProviderInfo
+from aiq.builder.workflow import Workflow
+from aiq.cli.type_registry import GlobalTypeRegistry
+from aiq.cli.type_registry import TypeRegistry
+from aiq.data_models.component import ComponentGroup
+from aiq.data_models.component_ref import EmbedderRef
+from aiq.data_models.component_ref import FunctionRef
+from aiq.data_models.component_ref import LLMRef
+from aiq.data_models.component_ref import MemoryRef
+from aiq.data_models.component_ref import RetrieverRef
+from aiq.data_models.config import AIQConfig
+from aiq.data_models.config import GeneralConfig
+from aiq.data_models.embedder import EmbedderBaseConfig
+from aiq.data_models.function import FunctionBaseConfig
+from aiq.data_models.function_dependencies import FunctionDependencies
+from aiq.data_models.llm import LLMBaseConfig
+from aiq.data_models.memory import MemoryBaseConfig
+from aiq.data_models.retriever import RetrieverBaseConfig
+from aiq.data_models.telemetry_exporter import TelemetryExporterBaseConfig
+from aiq.memory.interfaces import MemoryEditor
+from aiq.profiler.decorators.framework_wrapper import chain_wrapped_build_fn
+from aiq.profiler.utils import detect_llm_frameworks_in_build_fn
+from aiq.utils.type_utils import override
+
+logger = logging.getLogger(__name__)
+
+
+@dataclasses.dataclass
+class ConfiguredExporter:
+    config: TelemetryExporterBaseConfig
+    instance: SpanExporter
+
+
+@dataclasses.dataclass
+class ConfiguredFunction:
+    config: FunctionBaseConfig
+    instance: Function
+
+
+@dataclasses.dataclass
+class ConfiguredLLM:
+    config: LLMBaseConfig
+    instance: LLMProviderInfo
+
+
+@dataclasses.dataclass
+class ConfiguredEmbedder:
+    config: EmbedderBaseConfig
+    instance: EmbedderProviderInfo
+
+
+@dataclasses.dataclass
+class ConfiguredMemory:
+    config: MemoryBaseConfig
+    instance: MemoryEditor
+
+
+@dataclasses.dataclass
+class ConfiguredRetriever:
+    config: RetrieverBaseConfig
+    instance: RetrieverProviderInfo
+
+
+# pylint: disable=too-many-public-methods
+class WorkflowBuilder(Builder, AbstractAsyncContextManager):
+
+    def __init__(self, *, general_config: GeneralConfig | None = None, registry: TypeRegistry | None = None):
+
+        if general_config is None:
+            general_config = GeneralConfig()
+
+        if registry is None:
+            registry = GlobalTypeRegistry.get()
+
+        self.general_config = general_config
+
+        self._registry = registry
+
+        self._logging_handlers: dict[str, logging.Handler] = {}
+        self._exporters: dict[str, ConfiguredExporter] = {}
+
+        self._functions: dict[str, ConfiguredFunction] = {}
+        self._workflow: ConfiguredFunction | None = None
+
+        self._llms: dict[str, ConfiguredLLM] = {}
+        self._embedders: dict[str, ConfiguredEmbedder] = {}
+        self._memory_clients: dict[str, ConfiguredMemory] = {}
+        self._retrievers: dict[str, ConfiguredRetriever] = {}
+
+        self._context_state = AIQContextState.get()
+
+        self._exit_stack: AsyncExitStack | None = None
+
+        # Create a mapping to track function name -> other function names it depends on
+        self.function_dependencies: dict[str, FunctionDependencies] = {}
+        self.current_function_building: str | None = None
+
+    async def __aenter__(self):
+
+        self._exit_stack = AsyncExitStack()
+
+        # Get the exporter info from the config
+        telemetry_config = self.general_config.telemetry
+
+        for key, logging_config in telemetry_config.logging.items():
+            # Use the same pattern as tracing, but for logging
+            logging_info = self._registry.get_logging_method(type(logging_config))
+            handler = await self._exit_stack.enter_async_context(logging_info.build_fn(logging_config, self))
+
+            # Type check
+            if not isinstance(handler, logging.Handler):
+                raise TypeError(f"Expected a logging.Handler from {key}, got {type(handler)}")
+
+            # Store them in a dict so we can un-register them if needed
+            self._logging_handlers[key] = handler
+
+            # Now attach to AgentIQ's root logger
+            logging.getLogger().addHandler(handler)
+
+        provider = TracerProvider()
+        trace.set_tracer_provider(provider)
+
+        for key, trace_exporter_config in telemetry_config.tracing.items():
+
+            exporter_info = self._registry.get_telemetry_exporter(type(trace_exporter_config))
+
+            instance = await self._exit_stack.enter_async_context(exporter_info.build_fn(trace_exporter_config, self))
+
+            span_processor_instance = BatchSpanProcessor(instance)
+            provider.add_span_processor(span_processor_instance)
+
+            self._exporters[key] = ConfiguredExporter(config=trace_exporter_config, instance=instance)
+
+        return self
+
+    async def __aexit__(self, *exc_details):
+
+        assert self._exit_stack is not None, "Exit stack not initialized"
+
+        for _, handler in self._logging_handlers.items():
+            logging.getLogger().removeHandler(handler)
+
+        await self._exit_stack.__aexit__(*exc_details)
+
+    def build(self, entry_function: str | None = None) -> Workflow:
+        """
+        Creates an instance of a workflow object using the added components and the desired entry function.
+
+        Parameters
+        ----------
+        entry_function : str | None, optional
+            The function name to use as the entry point for the created workflow. If None, the entry point will be the
+            specified workflow function. By default None
+
+        Returns
+        -------
+        Workflow
+            A created workflow.
+
+        Raises
+        ------
+        ValueError
+            If the workflow has not been set before building.
+        """
+
+        if (self._workflow is None):
+            raise ValueError("Must set a workflow before building")
+
+        # Build the config from the added objects
+        config = AIQConfig(general=self.general_config,
+                           functions={
+                               k: v.config
+                               for k, v in self._functions.items()
+                           },
+                           workflow=self._workflow.config,
+                           llms={
+                               k: v.config
+                               for k, v in self._llms.items()
+                           },
+                           embedders={
+                               k: v.config
+                               for k, v in self._embedders.items()
+                           },
+                           memory={
+                               k: v.config
+                               for k, v in self._memory_clients.items()
+                           },
+                           retrievers={
+                               k: v.config
+                               for k, v in self._retrievers.items()
+                           })
+
+        if (entry_function is None):
+            entry_fn_obj = self.get_workflow()
+        else:
+            entry_fn_obj = self.get_function(entry_function)
+
+        workflow = Workflow.from_entry_fn(config=config,
+                                          entry_fn=entry_fn_obj,
+                                          functions={
+                                              k: v.instance
+                                              for k, v in self._functions.items()
+                                          },
+                                          llms={
+                                              k: v.instance
+                                              for k, v in self._llms.items()
+                                          },
+                                          embeddings={
+                                              k: v.instance
+                                              for k, v in self._embedders.items()
+                                          },
+                                          memory={
+                                              k: v.instance
+                                              for k, v in self._memory_clients.items()
+                                          },
+                                          exporters={
+                                              k: v.instance
+                                              for k, v in self._exporters.items()
+                                          },
+                                          retrievers={
+                                              k: v.instance
+                                              for k, v in self._retrievers.items()
+                                          },
+                                          context_state=self._context_state)
+
+        return workflow
+
+    def _get_exit_stack(self) -> AsyncExitStack:
+
+        if self._exit_stack is None:
+            raise ValueError(
+                "Exit stack not initialized. Did you forget to call `async with WorkflowBuilder() as builder`?")
+
+        return self._exit_stack
+
+    async def _build_function(self, name: str, config: FunctionBaseConfig) -> ConfiguredFunction:
+        registration = self._registry.get_function(type(config))
+
+        inner_builder = ChildBuilder(self)
+
+        # We need to do this for every function because we don't know
+        # Where LLama Index Agents are Instantiated and Settings need to
+        # be set before the function is built
+        # It's only slower the first time because of the import
+        # So we can afford to do this for every function
+
+        llms = {k: v.instance for k, v in self._llms.items()}
+        function_frameworks = detect_llm_frameworks_in_build_fn(registration)
+
+        build_fn = chain_wrapped_build_fn(registration.build_fn, llms, function_frameworks)
+
+        # Set the currently building function so the ChildBuilder can track dependencies
+        self.current_function_building = config.type
+        # Empty set of dependencies for the current function
+        self.function_dependencies[config.type] = FunctionDependencies()
+
+        build_result = await self._get_exit_stack().enter_async_context(build_fn(config, inner_builder))
+
+        self.function_dependencies[name] = inner_builder.dependencies
+
+        # If the build result is a function, wrap it in a FunctionInfo
+        if inspect.isfunction(build_result):
+
+            build_result = FunctionInfo.from_fn(build_result)
+
+        if (isinstance(build_result, FunctionInfo)):
+            # Create the function object
+            build_result = LambdaFunction.from_info(config=config, info=build_result)
+
+        if (not isinstance(build_result, Function)):
+            raise ValueError("Expected a function, FunctionInfo object, or FunctionBase object to be "
+                             f"returned from the function builder. Got {type(build_result)}")
+
+        return ConfiguredFunction(config=config, instance=build_result)
+
+    @override
+    async def add_function(self, name: str | FunctionRef, config: FunctionBaseConfig) -> Function:
+
+        if (name in self._functions):
+            raise ValueError(f"Function `{name}` already exists in the list of functions")
+
+        build_result = await self._build_function(name=name, config=config)
+
+        self._functions[name] = build_result
+
+        return build_result.instance
+
+    @override
+    def get_function(self, name: str | FunctionRef) -> Function:
+
+        if name not in self._functions:
+            raise ValueError(f"Function `{name}` not found")
+
+        return self._functions[name].instance
+
+    @override
+    def get_function_config(self, name: str | FunctionRef) -> FunctionBaseConfig:
+        if name not in self._functions:
+            raise ValueError(f"Function `{name}` not found")
+
+        return self._functions[name].config
+
+    @override
+    async def set_workflow(self, config: FunctionBaseConfig) -> Function:
+
+        if self._workflow is not None:
+            warnings.warn("Overwriting existing workflow")
+
+        build_result = await self._build_function(name="<workflow>", config=config)
+
+        self._workflow = build_result
+
+        return build_result.instance
+
+    @override
+    def get_workflow(self) -> Function:
+
+        if self._workflow is None:
+            raise ValueError("No workflow set")
+
+        return self._workflow.instance
+
+    @override
+    def get_workflow_config(self) -> FunctionBaseConfig:
+        if self._workflow is None:
+            raise ValueError("No workflow set")
+
+        return self._workflow.config
+
+    @override
+    def get_function_dependencies(self, fn_name: str | FunctionRef) -> FunctionDependencies:
+        return self.function_dependencies[fn_name]
+
+    @override
+    def get_tool(self, fn_name: str | FunctionRef, wrapper_type: LLMFrameworkEnum | str):
+
+        if fn_name not in self._functions:
+            raise ValueError(f"Function `{fn_name}` not found in list of functions")
+
+        fn = self._functions[fn_name]
+
+        try:
+            # Using the registry, get the tool wrapper for the requested framework
+            tool_wrapper_reg = self._registry.get_tool_wrapper(llm_framework=wrapper_type)
+
+            # Wrap in the correct wrapper
+            return tool_wrapper_reg.build_fn(fn_name, fn.instance, self)
+        except Exception as e:
+            logger.error("Error fetching tool `%s`", fn_name, exc_info=True)
+            raise e
+
+    @override
+    async def add_llm(self, name: str | LLMRef, config: LLMBaseConfig):
+
+        if (name in self._llms):
+            raise ValueError(f"LLM `{name}` already exists in the list of LLMs")
+
+        try:
+            llm_info = self._registry.get_llm_provider(type(config))
+
+            info_obj = await self._get_exit_stack().enter_async_context(llm_info.build_fn(config, self))
+
+            self._llms[name] = ConfiguredLLM(config=config, instance=info_obj)
+        except Exception as e:
+            logger.error("Error adding llm `%s` with config `%s`", name, config, exc_info=True)
+            raise e
+
+    @override
+    async def get_llm(self, llm_name: str | LLMRef, wrapper_type: LLMFrameworkEnum | str):
+
+        if (llm_name not in self._llms):
+            raise ValueError(f"LLM `{llm_name}` not found")
+
+        try:
+            # Get llm info
+            llm_info = self._llms[llm_name]
+
+            # Generate wrapped client from registered client info
+            client_info = self._registry.get_llm_client(config_type=type(llm_info.config), wrapper_type=wrapper_type)
+
+            client = await self._get_exit_stack().enter_async_context(client_info.build_fn(llm_info.config, self))
+
+            # Return a frameworks specific client
+            return client
+        except Exception as e:
+            logger.error("Error getting llm `%s` with wrapper `%s`", llm_name, wrapper_type, exc_info=True)
+            raise e
+
+    @override
+    def get_llm_config(self, llm_name: str | LLMRef) -> LLMBaseConfig:
+
+        if llm_name not in self._llms:
+            raise ValueError(f"LLM `{llm_name}` not found")
+
+        # Return the tool configuration object
+        return self._llms[llm_name].config
+
+    @override
+    async def add_embedder(self, name: str | EmbedderRef, config: EmbedderBaseConfig):
+
+        if (name in self._embedders):
+            raise ValueError(f"Embedder `{name}` already exists in the list of embedders")
+
+        try:
+            embedder_info = self._registry.get_embedder_provider(type(config))
+
+            info_obj = await self._get_exit_stack().enter_async_context(embedder_info.build_fn(config, self))
+
+            self._embedders[name] = ConfiguredEmbedder(config=config, instance=info_obj)
+        except Exception as e:
+            logger.error("Error adding embedder `%s` with config `%s`", name, config, exc_info=True)
+
+            raise e
+
+    @override
+    async def get_embedder(self, embedder_name: str | EmbedderRef, wrapper_type: LLMFrameworkEnum | str):
+
+        if (embedder_name not in self._embedders):
+            raise ValueError(f"Embedder `{embedder_name}` not found")
+
+        try:
+            # Get embedder info
+            embedder_info = self._embedders[embedder_name]
+
+            # Generate wrapped client from registered client info
+            client_info = self._registry.get_embedder_client(config_type=type(embedder_info.config),
+                                                             wrapper_type=wrapper_type)
+            client = await self._get_exit_stack().enter_async_context(client_info.build_fn(embedder_info.config, self))
+
+            # Return a frameworks specific client
+            return client
+        except Exception as e:
+            logger.error("Error getting embedder `%s` with wrapper `%s`", embedder_name, wrapper_type, exc_info=True)
+            raise e
+
+    @override
+    def get_embedder_config(self, embedder_name: str | EmbedderRef) -> EmbedderBaseConfig:
+
+        if embedder_name not in self._embedders:
+            raise ValueError(f"Tool `{embedder_name}` not found")
+
+        # Return the tool configuration object
+        return self._embedders[embedder_name].config
+
+    @override
+    async def add_memory_client(self, name: str | MemoryRef, config: MemoryBaseConfig) -> MemoryEditor:
+
+        if (name in self._memory_clients):
+            raise ValueError(f"Memory `{name}` already exists in the list of memories")
+
+        memory_info = self._registry.get_memory(type(config))
+
+        info_obj = await self._get_exit_stack().enter_async_context(memory_info.build_fn(config, self))
+
+        self._memory_clients[name] = ConfiguredMemory(config=config, instance=info_obj)
+
+        return info_obj
+
+    @override
+    def get_memory_client(self, memory_name: str | MemoryRef) -> MemoryEditor:
+        """
+        Return the instantiated memory client for the given name.
+        """
+        if memory_name not in self._memory_clients:
+            raise ValueError(f"Memory `{memory_name}` not found")
+
+        return self._memory_clients[memory_name].instance
+
+    @override
+    def get_memory_client_config(self, memory_name: str | MemoryRef) -> MemoryBaseConfig:
+
+        if memory_name not in self._memory_clients:
+            raise ValueError(f"Memory `{memory_name}` not found")
+
+        # Return the tool configuration object
+        return self._memory_clients[memory_name].config
+
+    @override
+    async def add_retriever(self, name: str | RetrieverRef, config: RetrieverBaseConfig):
+
+        if (name in self._retrievers):
+            raise ValueError(f"Retriever '{name}' already exists in the list of retrievers")
+
+        try:
+            retriever_info = self._registry.get_retriever_provider(type(config))
+
+            info_obj = await self._get_exit_stack().enter_async_context(retriever_info.build_fn(config, self))
+
+            self._retrievers[name] = ConfiguredRetriever(config=config, instance=info_obj)
+
+        except Exception as e:
+            logger.error("Error adding retriever `%s` with config `%s`", name, config, exc_info=True)
+
+            raise e
+
+        # return info_obj
+
+    @override
+    async def get_retriever(self,
+                            retriever_name: str | RetrieverRef,
+                            wrapper_type: LLMFrameworkEnum | str | None = None):
+
+        if retriever_name not in self._retrievers:
+            raise ValueError(f"Retriever '{retriever_name}' not found")
+
+        try:
+            # Get retriever info
+            retriever_info = self._retrievers[retriever_name]
+
+            # Generate wrapped client from registered client info
+            client_info = self._registry.get_retriever_client(config_type=type(retriever_info.config),
+                                                              wrapper_type=wrapper_type)
+
+            client = await self._get_exit_stack().enter_async_context(client_info.build_fn(retriever_info.config, self))
+
+            # Return a frameworks specific client
+            return client
+        except Exception as e:
+            logger.error("Error getting retriever `%s` with wrapper `%s`", retriever_name, wrapper_type, exc_info=True)
+            raise e
+
+    @override
+    async def get_retriever_config(self, retriever_name: str | RetrieverRef) -> RetrieverBaseConfig:
+
+        if retriever_name not in self._retrievers:
+            raise ValueError(f"Retriever `{retriever_name}` not found")
+
+        return self._retrievers[retriever_name].config
+
+    @override
+    def get_user_manager(self):
+        return UserManagerHolder(context=AIQContext(self._context_state))
+
+    async def populate_builder(self, config: AIQConfig):
+
+        # Generate the build sequence
+        build_sequence = build_dependency_sequence(config)
+
+        # Loop over all objects and add to the workflow builder
+        for component_instance in build_sequence:
+            # Instantiate a the llm
+            if component_instance.component_group == ComponentGroup.LLMS:
+                await self.add_llm(component_instance.name, component_instance.config)
+            # Instantiate a the embedder
+            elif component_instance.component_group == ComponentGroup.EMBEDDERS:
+                await self.add_embedder(component_instance.name, component_instance.config)
+            # Instantiate a memory client
+            elif component_instance.component_group == ComponentGroup.MEMORY:
+                await self.add_memory_client(component_instance.name, component_instance.config)
+            # Instantiate a retriever client
+            elif component_instance.component_group == ComponentGroup.RETRIEVERS:
+                await self.add_retriever(component_instance.name, component_instance.config)
+            # Instantiate a function
+            elif component_instance.component_group == ComponentGroup.FUNCTIONS:
+                # If the function is the root, set it as the workflow later
+                if (not component_instance.is_root):
+                    await self.add_function(component_instance.name, component_instance.config)
+            else:
+                raise ValueError(f"Unknown component group {component_instance.component_group}")
+
+        # Instantiate the workflow
+        await self.set_workflow(config.workflow)
+
+    @classmethod
+    @asynccontextmanager
+    async def from_config(cls, config: AIQConfig):
+
+        async with cls(general_config=config.general) as builder:
+            await builder.populate_builder(config)
+            yield builder
+
+
+class ChildBuilder(Builder):
+
+    def __init__(self, workflow_builder: WorkflowBuilder) -> None:
+
+        self._workflow_builder = workflow_builder
+
+        self._dependencies = FunctionDependencies()
+
+    @property
+    def dependencies(self) -> FunctionDependencies:
+        return self._dependencies
+
+    @override
+    async def add_function(self, name: str, config: FunctionBaseConfig) -> Function:
+        return await self._workflow_builder.add_function(name, config)
+
+    @override
+    def get_function(self, name: str) -> Function:
+        # If a function tries to get another function, we assume it uses it
+        fn = self._workflow_builder.get_function(name)
+
+        self._dependencies.add_function(name)
+
+        return fn
+
+    @override
+    def get_function_config(self, name: str) -> FunctionBaseConfig:
+        return self._workflow_builder.get_function_config(name)
+
+    @override
+    async def set_workflow(self, config: FunctionBaseConfig) -> Function:
+        return await self._workflow_builder.set_workflow(config)
+
+    @override
+    def get_workflow(self) -> Function:
+        return self._workflow_builder.get_workflow()
+
+    @override
+    def get_workflow_config(self) -> FunctionBaseConfig:
+        return self._workflow_builder.get_workflow_config()
+
+    @override
+    def get_tool(self, fn_name: str, wrapper_type: LLMFrameworkEnum | str):
+        # If a function tries to get another function as a tool, we assume it uses it
+        fn = self._workflow_builder.get_tool(fn_name, wrapper_type)
+
+        self._dependencies.add_function(fn_name)
+
+        return fn
+
+    @override
+    async def add_llm(self, name: str, config: LLMBaseConfig):
+        return await self._workflow_builder.add_llm(name, config)
+
+    @override
+    async def get_llm(self, llm_name: str, wrapper_type: LLMFrameworkEnum | str):
+        llm = await self._workflow_builder.get_llm(llm_name, wrapper_type)
+
+        self._dependencies.add_llm(llm_name)
+
+        return llm
+
+    @override
+    def get_llm_config(self, llm_name: str) -> LLMBaseConfig:
+        return self._workflow_builder.get_llm_config(llm_name)
+
+    @override
+    async def add_embedder(self, name: str, config: EmbedderBaseConfig):
+        return await self._workflow_builder.add_embedder(name, config)
+
+    @override
+    async def get_embedder(self, embedder_name: str, wrapper_type: LLMFrameworkEnum | str):
+        embedder = await self._workflow_builder.get_embedder(embedder_name, wrapper_type)
+
+        self._dependencies.add_embedder(embedder_name)
+
+        return embedder
+
+    @override
+    def get_embedder_config(self, embedder_name: str) -> EmbedderBaseConfig:
+        return self._workflow_builder.get_embedder_config(embedder_name)
+
+    @override
+    async def add_memory_client(self, name: str, config: MemoryBaseConfig) -> MemoryEditor:
+        return await self._workflow_builder.add_memory_client(name, config)
+
+    @override
+    def get_memory_client(self, memory_name: str) -> MemoryEditor:
+        """
+        Return the instantiated memory client for the given name.
+        """
+        memory_client = self._workflow_builder.get_memory_client(memory_name)
+
+        self._dependencies.add_memory_client(memory_name)
+
+        return memory_client
+
+    @override
+    def get_memory_client_config(self, memory_name: str) -> MemoryBaseConfig:
+        return self._workflow_builder.get_memory_client_config(memory_name=memory_name)
+
+    @override
+    async def add_retriever(self, name: str, config: RetrieverBaseConfig):
+        return await self._workflow_builder.add_retriever(name, config)
+
+    @override
+    async def get_retriever(self, retriever_name: str, wrapper_type: LLMFrameworkEnum | str | None = None):
+        if not wrapper_type:
+            return await self._workflow_builder.get_retriever(retriever_name=retriever_name)
+        return await self._workflow_builder.get_retriever(retriever_name=retriever_name, wrapper_type=wrapper_type)
+
+    @override
+    async def get_retriever_config(self, retriever_name: str) -> RetrieverBaseConfig:
+        return await self._workflow_builder.get_retriever_config(retriever_name=retriever_name)
+
+    @override
+    def get_user_manager(self) -> UserManagerHolder:
+        return self._workflow_builder.get_user_manager()
+
+    @override
+    def get_function_dependencies(self, fn_name: str) -> FunctionDependencies:
+        return self._workflow_builder.get_function_dependencies(fn_name)