feat(workflow):

1. add list operator node for filtering, sorting, limiting, and extracting list items；
2. Increase the session variable to the "file" type

api/app/core/models/embedding.py

@@ -25,8 +25,34 @@ class RedBearEmbeddings(Embeddings):
     def _create_model(self, config: RedBearModelConfig) -> Embeddings:
         """根据配置创建 LangChain 模型"""
         embedding_class = get_provider_embedding_class(config.provider)
-        model_params = RedBearModelFactory.get_model_params(config)
-        return embedding_class(**model_params)
+        provider = config.provider.lower()
+        # Embedding models only need connection params, never LLM-specific ones
+        # (e.g. enable_thinking, model_kwargs) — build params directly.
+        if provider in [ModelProvider.OPENAI, ModelProvider.XINFERENCE, ModelProvider.GPUSTACK]:
+            import httpx
+            params = {
+                "model": config.model_name,
+                "base_url": config.base_url,
+                "api_key": config.api_key,
+                "timeout": httpx.Timeout(timeout=config.timeout, connect=60.0),
+                "max_retries": config.max_retries,
+            }
+        elif provider == ModelProvider.DASHSCOPE:
+            params = {
+                "model": config.model_name,
+                "dashscope_api_key": config.api_key,
+                "max_retries": config.max_retries,
+            }
+        elif provider == ModelProvider.OLLAMA:
+            params = {
+                "model": config.model_name,
+                "base_url": config.base_url,
+            }
+        elif provider == ModelProvider.BEDROCK:
+            params = RedBearModelFactory.get_model_params(config)
+        else:
+            params = RedBearModelFactory.get_model_params(config)
+        return embedding_class(**params)
     
     def _create_volcano_client(self, config: RedBearModelConfig):
         """创建火山引擎客户端"""

api/app/core/models/volcano_chat.py

@@ -6,14 +6,28 @@ ChatOpenAI 在解析流式 SSE 时只取 delta.content，会丢弃 delta.reasoni
 """
 from __future__ import annotations
 
-from typing import Any, Optional
+from typing import Any, Optional, Union
 
-from langchain_core.outputs import ChatGenerationChunk
+from langchain_core.outputs import ChatGenerationChunk, ChatResult
 from langchain_openai import ChatOpenAI
 
 
 class VolcanoChatOpenAI(ChatOpenAI):
-    """火山引擎 Chat 模型，支持深度思考内容（reasoning_content）的流式透传。"""
+    """火山引擎 Chat 模型，支持深度思考内容（reasoning_content）的流式和非流式透传。"""
+
+    def _create_chat_result(self, response: Union[dict, Any], generation_info: Optional[dict] = None) -> ChatResult:
+        result = super()._create_chat_result(response, generation_info)
+        # 将非流式响应中的 reasoning_content 补入 additional_kwargs
+        choices = response.choices if hasattr(response, "choices") else response.get("choices", [])
+        if choices:
+            message = choices[0].message if hasattr(choices[0], "message") else choices[0].get("message", {})
+            reasoning = (
+                getattr(message, "reasoning_content", None)
+                or (message.get("reasoning_content") if isinstance(message, dict) else None)
+            )
+            if reasoning and result.generations:
+                result.generations[0].message.additional_kwargs["reasoning_content"] = reasoning
+        return result
 
     def _convert_chunk_to_generation_chunk(
         self,