refactor(api): replace legacy agent runners with StrategyFactory in AgentChatAppRunner (Phase 4)

Replace the hardcoded FunctionCallAgentRunner / CotChatAgentRunner /
CotCompletionAgentRunner selection in AgentChatAppRunner with the new
AgentAppRunner class that uses StrategyFactory from Phase 1.

Before: AgentChatAppRunner manually selects FC/CoT runner class based on
model features and LLM mode, then instantiates it directly.

After: AgentChatAppRunner instantiates AgentAppRunner (from sandbox branch),
which internally uses StrategyFactory.create_strategy() to auto-select
the right strategy, and uses ToolInvokeHook for proper agent_invoke
with file handling and thought persistence.

This unifies the agent execution engine: both the new Agent V2 workflow
node and the legacy agent-chat app now use the same StrategyFactory
and AgentPattern implementations.

Also fix: command and file_upload nodes use string node_type instead of
BuiltinNodeTypes.COMMAND/FILE_UPLOAD (not in current graphon version).

46 tests pass. Flask starts successfully.

Made-with: Cursor

api/core/app/apps/agent_chat/app_runner.py

@@ -1,15 +1,12 @@
 import logging
 from typing import cast
 
-from graphon.model_runtime.entities.llm_entities import LLMMode
-from graphon.model_runtime.entities.model_entities import ModelFeature, ModelPropertyKey
+from graphon.model_runtime.entities.model_entities import ModelFeature
 from graphon.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from sqlalchemy import select
 
-from core.agent.cot_chat_agent_runner import CotChatAgentRunner
-from core.agent.cot_completion_agent_runner import CotCompletionAgentRunner
+from core.agent.agent_app_runner import AgentAppRunner
 from core.agent.entities import AgentEntity
-from core.agent.fc_agent_runner import FunctionCallAgentRunner
 from core.app.apps.agent_chat.app_config_manager import AgentChatAppConfig
 from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
 from core.app.apps.base_app_runner import AppRunner
@@ -194,22 +191,7 @@ class AgentChatAppRunner(AppRunner):
             raise ValueError("Message not found")
         db.session.close()
 
-        runner_cls: type[FunctionCallAgentRunner] | type[CotChatAgentRunner] | type[CotCompletionAgentRunner]
-        # start agent runner
-        if agent_entity.strategy == AgentEntity.Strategy.CHAIN_OF_THOUGHT:
-            # check LLM mode
-            if model_schema.model_properties.get(ModelPropertyKey.MODE) == LLMMode.CHAT:
-                runner_cls = CotChatAgentRunner
-            elif model_schema.model_properties.get(ModelPropertyKey.MODE) == LLMMode.COMPLETION:
-                runner_cls = CotCompletionAgentRunner
-            else:
-                raise ValueError(f"Invalid LLM mode: {model_schema.model_properties.get(ModelPropertyKey.MODE)}")
-        elif agent_entity.strategy == AgentEntity.Strategy.FUNCTION_CALLING:
-            runner_cls = FunctionCallAgentRunner
-        else:
-            raise ValueError(f"Invalid agent strategy: {agent_entity.strategy}")
-
-        runner = runner_cls(
+        runner = AgentAppRunner(
             tenant_id=app_config.tenant_id,
             application_generate_entity=application_generate_entity,
             conversation=conversation_result,

api/core/workflow/nodes/command/node.py

@@ -22,7 +22,7 @@ COMMAND_NODE_TIMEOUT_SECONDS = 60 * 10
 
 
 class CommandNode(Node[CommandNodeData]):
-    node_type = BuiltinNodeTypes.COMMAND
+    node_type = "command"
 
     def _render_template(self, template: str) -> str:
         parser = VariableTemplateParser(template=template)

api/core/workflow/nodes/file_upload/node.py

@@ -29,7 +29,7 @@ class FileUploadNode(Node[FileUploadNodeData]):
     files, it generates storage-backed presigned URLs and lets sandbox download directly.
     """
 
-    node_type = BuiltinNodeTypes.FILE_UPLOAD
+    node_type = "file-upload"
 
     @classmethod
     def version(cls) -> str: