Merge remote-tracking branch 'origin/develop' into refactor/memory-config-management

2025-12-24 16:07:27 +08:00
parent 0fff23d2a2 3afe547559
commit 292fef0fa7
24 changed files with 365 additions and 202 deletions
--- a/api/app/core/memory/agent/mcp_server/server.py
+++ b/api/app/core/memory/agent/mcp_server/server.py
@@ -131,6 +131,18 @@ def main():
        
        # Get MCP port from environment (default: 8081)
        mcp_port = int(os.getenv("MCP_PORT", "8081"))
+        logger.info(f"Starting MCP server on {settings.SERVER_IP}:{mcp_port} with SSE transport")
+        
+        # Configure DNS rebinding protection for Docker container compatibility
+        from mcp.server.fastmcp.server import TransportSecuritySettings
+        
+        # Disable DNS rebinding protection to allow Docker container hostnames
+        # This allows containers to connect using service names like 'mcp-server'
+        mcp.settings.transport_security = TransportSecuritySettings(
+            enable_dns_rebinding_protection=False,
+        )
+        logger.info("DNS rebinding protection: disabled for Docker container compatibility")
+        
        # logger.info(f"Starting MCP server on {settings.SERVER_IP}:{mcp_port} with SSE transport")
        
        # Run the server with SSE transport for HTTP connections
--- a/api/app/core/memory/storage_services/reflection_engine/example/example.json
+++ b/api/app/core/memory/storage_services/reflection_engine/example/example.json
@@ -50,9 +50,7 @@
        "entity2_name": "用户",
        "entity2": {
          "description": "叙述者，讲述个人工作与生活经历的个体",
-          "statement_id": "62beac695b1346f4871740a45db88782",
-          "name": "用户",
-          "id": "3d3896797b334572a80d57590026063d"
+          "name": "用户"
        }
      },
      {
@@ -62,9 +60,7 @@
        "entity2_name": "身份信息",
        "entity2": {
          "description": "用于个人身份识别的数据",
-          "statement_id": "030afd362e9b4110b139e68e5d3e7143",
-          "name": "身份信息",
-          "id": "aa766a517e82490599a9b3af54cfd933"
+          "name": "身份信息"
        }
      },
      {
@@ -74,9 +70,7 @@
        "entity2_name": "6222023847595898",
        "entity2": {
          "description": "用户的银行卡号码",
-          "statement_id": "6c7567cd1f3c478bb42d1b65383e6f2f",
-          "name": "6222023847595898",
-          "id": "610ba361918f4e68a65ce6ad06e5c7a0"
+          "name": "6222023847595898"
        }
      },
      {
@@ -88,9 +82,7 @@
          "entity_idx": 1,
          "aliases": ["上海办"],
          "description": "位于上海的工作办公场所",
-          "statement_id": "8b1b12e23b844b8088dfeb67da6ad669",
-          "name": "上海办公室",
-          "id": "fb702ef695c14e14af3e56786bc8815b"
+          "name": "上海办公室"
        }
      },
      {
@@ -101,9 +93,7 @@
        "entity2": {
          "aliases": ["京", "京城", "北平"],
          "description": "中国的首都城市，用户主要工作和生活所在地",
-          "statement_id": "62beac695b1346f4871740a45db88782",
-          "name": "北京",
-          "id": "81b2d1a571bb46a08a2d7a1e87efb945"
+          "name": "北京"
        }
      },
      {
@@ -113,9 +103,7 @@
        "entity2_name": "身份证号",
        "entity2": {
          "description": "中华人民共和国公民的身份号码",
-          "statement_id": "030afd362e9b4110b139e68e5d3e7143",
-          "name": "身份证号",
-          "id": "3e5f920645b2404fadb0e9ff60d1306e"
+          "name": "身份证号"
        }
      }
    ]
--- a/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py
+++ b/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py
@@ -269,8 +269,6 @@ class ReflectionEngine:
            # # 检查是否真的有冲突
            conflicts_found=''

-            # 记录冲突数据
-            await self._log_data("conflict", conflict_data)
            conflicts_found=''
            # 3. 解决冲突
            solved_data = await self._resolve_conflicts(conflict_data, statement_databasets)
@@ -288,8 +286,6 @@ class ReflectionEngine:
            conflicts_resolved = len(solved_data)
            logging.info(f"解决了 {conflicts_resolved} 个冲突")

-            # 记录解决方案
-            await self._log_data("solved_data", solved_data)

            # 4. 应用反思结果（更新记忆库）
            memories_updated=await self._apply_reflection_results(solved_data)
@@ -390,14 +386,7 @@ class ReflectionEngine:
            memory_verifies.append(item['memory_verify'])
        result_data['memory_verifies'] = memory_verifies
        result_data['quality_assessments'] = quality_assessments
-
-        # 检查是否真的有冲突
-        has_conflict = conflict_data[0].get('conflict', False)
-        conflicts_found = len(conflict_data[0]['data']) if has_conflict else 0
-        logging.info(f"冲突状态: {has_conflict}, 发现 {conflicts_found} 个冲突")
-
-        # 记录冲突数据
-        await self._log_data("conflict", conflict_data)
+        conflicts_found=''

        # Clearn conflict_data，And memory_verify和quality_assessment
        cleaned_conflict_data = []
@@ -407,6 +396,7 @@ class ReflectionEngine:
                'conflict': item['conflict']
            }
            cleaned_conflict_data.append(cleaned_item)
+
        # 3. 解决冲突
        solved_data = await self._resolve_conflicts(cleaned_conflict_data, source_data)
        if not solved_data:
@@ -645,26 +635,7 @@ class ReflectionEngine:
        success_count = await neo4j_data(changes)
        return success_count

-    async def _log_data(self, label: str, data: Any) -> None:
-        """
-        记录数据到文件

-        Args:
-            label: 数据标签
-            data: 要记录的数据
-        """
-
-        def _write():
-            try:
-                with open("reflexion_data.json", "a", encoding="utf-8") as f:
-                    f.write(f"### {label} ###\n")
-                    json.dump(data, f, ensure_ascii=False, indent=4)
-                    f.write("\n\n")
-            except Exception as e:
-                logging.warning(f"记录数据失败: {e}")
-
-        # 在后台线程中执行写入，避免阻塞事件循环
-        await asyncio.to_thread(_write)

    # 基于时间的反思方法
    async def time_based_reflection(
@@ -753,4 +724,3 @@ class ReflectionEngine:
            raise ValueError(f"未知的反思基线: {self.config.baseline}")


-
--- a/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2
+++ b/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2
@@ -17,10 +17,12 @@
 - **日期属性冲突**: 同一人的生日等单值属性出现多值
 - **先后约束违反**: 存在A→B约束但t(A)>t(B)（如入学>毕业）
 - **互斥重叠**: 同一时间出现在不同地点等互斥事件
+- **隐私审核**: 存在隐私信息也作为冲突输出当{{ memory_verify }}是true的时候
 ### 事实冲突
 - **属性互斥**: 同一实体的相反属性（喜欢↔不喜欢）
 - **关系矛盾**: 同一实体在相同语境下的不同关系描述
 - **身份冲突**: 同一实体被赋予不同类型或角色
+- **隐私审核**: 存在隐私信息也作为冲突输出当{{ memory_verify }}是true的时候
 ### 混合冲突
 检测所有逻辑不一致或相互矛盾的记录。
 **检测原则**:
--- a/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2
+++ b/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2
@@ -171,7 +171,6 @@
  ]
 }
 ```
-
 **输出要求**:
 - 只输出JSON，不添加解释文本
 - 使用标准双引号，必要时转义
--- a/api/app/core/memory/utils/prompt/template_render.py
+++ b/api/app/core/memory/utils/prompt/template_render.py
@@ -7,7 +7,7 @@ from typing import List, Dict, Any
 prompt_dir = os.path.join(os.path.dirname(__file__), "prompts")
 prompt_env = Environment(loader=FileSystemLoader(prompt_dir))

-async def render_evaluate_prompt(evaluate_data: List[Any], schema: Dict[str, Any],
+async def render_evaluate_prompt(evaluate_data: List[Any], schema: Any,
                                 baseline: str = "TIME",
                                 memory_verify: bool = False,quality_assessment:bool = False,
                                 statement_databasets: List[str] = [],language_type:str = "zh") -> str:
@@ -16,7 +16,7 @@ async def render_evaluate_prompt(evaluate_data: List[Any], schema: Dict[str, Any

    Args:
        evaluate_data: The data to evaluate
-        schema: The JSON schema to use for the output.
+        schema: The Pydantic model class or JSON schema to use for the output.
        baseline: The baseline type for conflict detection (TIME/FACT/TIME-FACT)
        memory_verify: Whether to enable memory verification for privacy detection

@@ -25,9 +25,17 @@ async def render_evaluate_prompt(evaluate_data: List[Any], schema: Dict[str, Any
    """
    template = prompt_env.get_template("evaluate.jinja2")

+    # Convert Pydantic model to JSON schema if needed
+    if hasattr(schema, 'model_json_schema'):
+        json_schema = schema.model_json_schema()
+    elif hasattr(schema, 'schema'):
+        json_schema = schema.schema()
+    else:
+        json_schema = schema
+
    rendered_prompt = template.render(
        evaluate_data=evaluate_data,
-        json_schema=schema,
+        json_schema=json_schema,
        baseline=baseline,
        memory_verify=memory_verify,
        quality_assessment=quality_assessment,
@@ -36,14 +44,15 @@ async def render_evaluate_prompt(evaluate_data: List[Any], schema: Dict[str, Any
    )
    return rendered_prompt

-async def render_reflexion_prompt(data: Dict[str, Any], schema: Dict[str, Any], baseline: str, memory_verify: bool = False,
+
+async def render_reflexion_prompt(data: Dict[str, Any], schema: Any, baseline: str, memory_verify: bool = False,
                                  statement_databasets: List[str] = [],language_type:str = "zh") -> str:
    """
    Renders the reflexion prompt using the reflexion_optimized.jinja2 template.

    Args:
        data: The data to reflex on.
-        schema: The JSON schema to use for the output.
+        schema: The Pydantic model class or JSON schema to use for the output.
        baseline: The baseline type for conflict resolution.

    Returns:
@@ -51,7 +60,15 @@ async def render_reflexion_prompt(data: Dict[str, Any], schema: Dict[str, Any],
    """
    template = prompt_env.get_template("reflexion.jinja2")

-    rendered_prompt = template.render(data=data, json_schema=schema,
+    # Convert Pydantic model to JSON schema if needed
+    if hasattr(schema, 'model_json_schema'):
+        json_schema = schema.model_json_schema()
+    elif hasattr(schema, 'schema'):
+        json_schema = schema.schema()
+    else:
+        json_schema = schema
+
+    rendered_prompt = template.render(data=data, json_schema=json_schema,
                                      baseline=baseline,memory_verify=memory_verify,
                                      statement_databasets=statement_databasets,language_type=language_type)

--- a/api/app/core/rag/deepdoc/parser/mineru_parser.py
+++ b/api/app/core/rag/deepdoc/parser/mineru_parser.py
@@ -13,7 +13,7 @@ from io import BytesIO
 from os import PathLike
 from pathlib import Path
 from queue import Empty, Queue
-from typing import Any, Callable, Optional
+from typing import Any, Callable, Optional, Union

 import numpy as np
 import pdfplumber
@@ -439,7 +439,7 @@ class MinerUParser(RAGPdfParser):
    def parse_pdf(
        self,
        filepath: str | PathLike[str],
-        binary: BytesIO | bytes,
+        binary: Optional[Union[BytesIO, bytes]] = None,
        callback: Optional[Callable] = None,
        *,
        output_dir: Optional[str] = None,
--- a/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py
+++ b/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py
@@ -740,8 +740,9 @@ class ElasticSearchVector(BaseVector):
            self._client.indices.create(index=self._collection_name, body=index_mapping)


-class ElasticSearchVectorFactory(ABC):
-    def init_vector(self, knowledge: Knowledge) -> ElasticSearchVector:
+class ElasticSearchVectorFactory:
+    @staticmethod
+    def init_vector(knowledge: Knowledge) -> ElasticSearchVector:
        collection_name = f"Vector_index_{knowledge.id}_Node"

        # Use regular Elasticsearch with config values
@@ -763,17 +764,17 @@ class ElasticSearchVectorFactory(ABC):
            }
        )

-        if knowledge.embedding and knowledge.reranker:
-            return ElasticSearchVector(
-                index_name=collection_name,
-                config=ElasticSearchConfig(**config_dict),
-                embedding_config=knowledge.embedding.api_keys[0],
-                reranker_config=knowledge.reranker.api_keys[0]
-            )
-        else:
-            if knowledge.embedding is None:
-                raise ValueError(f"embedding_id config error: {str(knowledge.embedding_id)}")
-            if knowledge.reranker is None:
-                raise ValueError(f"reranker_id config error: {str(knowledge.reranker_id)}")
+        if knowledge.embedding is None:
+            raise ValueError(f"embedding_id config error: {str(knowledge.embedding_id)}")
+        if knowledge.reranker is None:
+            raise ValueError(f"reranker_id config error: {str(knowledge.reranker_id)}")
+
+        return ElasticSearchVector(
+            index_name=collection_name,
+            config=ElasticSearchConfig(**config_dict),
+            embedding_config=knowledge.embedding.api_keys[0],
+            reranker_config=knowledge.reranker.api_keys[0]
+        )
+


--- a/api/app/core/workflow/nodes/init.py
+++ b/api/app/core/workflow/nodes/init.py
@@ -9,7 +9,7 @@ from app.core.workflow.nodes.assigner import AssignerNode
 from app.core.workflow.nodes.base_node import BaseNode, WorkflowState
 from app.core.workflow.nodes.end import EndNode
 from app.core.workflow.nodes.if_else import IfElseNode
-# from app.core.workflow.nodes.knowledge import KnowledgeRetrievalNode
+from app.core.workflow.nodes.knowledge import KnowledgeRetrievalNode
 from app.core.workflow.nodes.llm import LLMNode
 from app.core.workflow.nodes.node_factory import NodeFactory, WorkflowNode
 from app.core.workflow.nodes.start import StartNode
@@ -26,6 +26,6 @@ __all__ = [
    "EndNode",
    "NodeFactory",
    "WorkflowNode",
-    # "KnowledgeRetrievalNode",
+    "KnowledgeRetrievalNode",
    "AssignerNode",
 ]
--- a/api/app/core/workflow/nodes/assigner/config.py
+++ b/api/app/core/workflow/nodes/assigner/config.py
@@ -1,21 +1,32 @@
-from pydantic import Field
+from pydantic import Field, BaseModel

 from app.core.workflow.nodes.base_config import BaseNodeConfig
 from app.core.workflow.nodes.enums import AssignmentOperator


-class AssignerNodeConfig(BaseNodeConfig):
+class AssignmentItem(BaseModel):
+    """
+    Single assignment definition.
+    """
+
    variable_selector: str | list[str] = Field(
        ...,
-        description="Variables to be assigned",
+        description="Target variable name(s) to assign",
    )

    operation: AssignmentOperator = Field(
        ...,
-        description="Operator to assign",
+        description="Assignment operator",
    )

    value: str | list[str] = Field(
        ...,
-        description="Values to assign",
+        description="Value(s) to assign to the variable(s)",
+    )
+
+
+class AssignerNodeConfig(BaseNodeConfig):
+    assignments: list[AssignmentItem] = Field(
+        ...,
+        description="List of variable assignment definitions",
    )
--- a/api/app/core/workflow/nodes/assigner/node.py
+++ b/api/app/core/workflow/nodes/assigner/node.py
@@ -29,52 +29,52 @@ class AssignerNode(BaseNode):
        """
        # Initialize a variable pool for accessing conversation, node, and system variables
        pool = VariablePool(state)
+        for assignment in self.typed_config.assignments:
+            # Get the target variable selector (e.g., "conv.test")
+            variable_selector = assignment.variable_selector
+            if isinstance(variable_selector, str):
+                # Support dot-separated string paths, e.g., "conv.test" -> ["conv", "test"]
+                variable_selector = variable_selector.split('.')

-        # Get the target variable selector (e.g., "conv.test")
-        variable_selector = self.typed_config.variable_selector
-        if isinstance(variable_selector, str):
-            # Support dot-separated string paths, e.g., "conv.test" -> ["conv", "test"]
-            variable_selector = variable_selector.split('.')
+            # Only conversation variables ('conv') are allowed
+            if variable_selector[0] != 'conv':  # TODO: Loop node variable support (Feature)
+                raise ValueError("Only conversation variables can be assigned.")

-        # Only conversation variables ('conv') are allowed
-        if variable_selector[0] != 'conv':  # TODO: Loop node variable support (Feature)
-            raise ValueError("Only conversation variables can be assigned.")
+            # Get the value or expression to assign
+            value = assignment.value
+            if isinstance(value, list):
+                value = '.'.join(value)
+            value = ExpressionEvaluator.evaluate(
+                expression=value,
+                variables=pool.get_all_conversation_vars(),
+                node_outputs=pool.get_all_node_outputs(),
+                system_vars=pool.get_all_system_vars(),
+            )

-        # Get the value or expression to assign
-        value = self.typed_config.value
-        if isinstance(value, list):
-            value = '.'.join(value)
-        value = ExpressionEvaluator.evaluate(
-            expression=value,
-            variables=pool.get_all_conversation_vars(),
-            node_outputs=pool.get_all_node_outputs(),
-            system_vars=pool.get_all_system_vars(),
-        )
+            # Select the appropriate assignment operator instance based on the target variable type
+            operator: AssignmentOperatorInstance = AssignmentOperator.get_operator(pool.get(variable_selector))(
+                pool, variable_selector, value
+            )

-        # Select the appropriate assignment operator instance based on the target variable type
-        operator: AssignmentOperatorInstance = AssignmentOperator.get_operator(pool.get(variable_selector))(
-            pool, variable_selector, value
-        )
-
-        # Execute the configured assignment operation
-        match self.typed_config.operation:
-            case AssignmentOperator.ASSIGN:
-                operator.assign()
-            case AssignmentOperator.CLEAR:
-                operator.clear()
-            case AssignmentOperator.ADD:
-                operator.add()
-            case AssignmentOperator.SUBTRACT:
-                operator.subtract()
-            case AssignmentOperator.MULTIPLY:
-                operator.multiply()
-            case AssignmentOperator.DIVIDE:
-                operator.divide()
-            case AssignmentOperator.APPEND:
-                operator.append()
-            case AssignmentOperator.REMOVE_FIRST:
-                operator.remove_first()
-            case AssignmentOperator.REMOVE_LAST:
-                operator.remove_last()
-            case _:
-                raise ValueError(f"Invalid Operator: {self.typed_config.operation}")
+            # Execute the configured assignment operation
+            match assignment.operation:
+                case AssignmentOperator.ASSIGN:
+                    operator.assign()
+                case AssignmentOperator.CLEAR:
+                    operator.clear()
+                case AssignmentOperator.ADD:
+                    operator.add()
+                case AssignmentOperator.SUBTRACT:
+                    operator.subtract()
+                case AssignmentOperator.MULTIPLY:
+                    operator.multiply()
+                case AssignmentOperator.DIVIDE:
+                    operator.divide()
+                case AssignmentOperator.APPEND:
+                    operator.append()
+                case AssignmentOperator.REMOVE_FIRST:
+                    operator.remove_first()
+                case AssignmentOperator.REMOVE_LAST:
+                    operator.remove_last()
+                case _:
+                    raise ValueError(f"Invalid Operator: {assignment.operation}")
--- a/api/app/core/workflow/nodes/configs.py
+++ b/api/app/core/workflow/nodes/configs.py
@@ -14,7 +14,7 @@ from app.core.workflow.nodes.llm.config import LLMNodeConfig, MessageConfig
 from app.core.workflow.nodes.agent.config import AgentNodeConfig
 from app.core.workflow.nodes.transform.config import TransformNodeConfig
 from app.core.workflow.nodes.if_else.config import IfElseNodeConfig
-# from app.core.workflow.nodes.knowledge.config import KnowledgeRetrievalNodeConfig
+from app.core.workflow.nodes.knowledge.config import KnowledgeRetrievalNodeConfig
 from app.core.workflow.nodes.assigner.config import AssignerNodeConfig

 __all__ = [
@@ -30,6 +30,6 @@ __all__ = [
    "AgentNodeConfig",
    "TransformNodeConfig",
    "IfElseNodeConfig",
-    # "KnowledgeRetrievalNodeConfig",
+    "KnowledgeRetrievalNodeConfig",
    "AssignerNodeConfig",
 ]
--- a/api/app/core/workflow/nodes/end/node.py
+++ b/api/app/core/workflow/nodes/end/node.py
@@ -9,28 +9,29 @@ import re
 import asyncio

 from app.core.workflow.nodes.base_node import BaseNode, WorkflowState
+from app.core.workflow.nodes.enums import NodeType

 logger = logging.getLogger(__name__)


 class EndNode(BaseNode):
    """End 节点
-    
+
    工作流的结束节点，根据配置的模板输出最终结果。
    支持实时流式输出：如果模板引用了上游节点的输出，会实时监听其流式缓冲区。
    """
-    
+
    async def execute(self, state: WorkflowState) -> str:
        """执行 end 节点业务逻辑
-        
+
        Args:
            state: 工作流状态
-        
+
        Returns:
            最终输出字符串
        """
        logger.info(f"节点 {self.node_id} (End) 开始执行")
-        
+
        # 获取配置的输出模板
        output_template = self.config.get("output")

@@ -39,11 +40,11 @@ class EndNode(BaseNode):
            output = self._render_template(output_template, state)
        else:
            output = "工作流已完成"
-        
+
        # 统计信息（用于日志）
        node_outputs = state.get("node_outputs", {})
        total_nodes = len(node_outputs)
-        
+
        logger.info(f"节点 {self.node_id} (End) 执行完成，共执行 {total_nodes} 个节点")

        return output
@@ -127,24 +128,26 @@ class EndNode(BaseNode):
        return parts

    async def execute_stream(self, state: WorkflowState):
-        """流式执行 end 节点业务逻辑
+        """Execute End node business logic (streaming)

-        智能输出策略：
-        1. 检测模板中是否引用了直接上游节点
-        2. 如果引用了，只输出该引用**之后**的部分（后缀）
-        3. 前缀和引用内容已经在上游节点流式输出时发送了
+        Smart output strategy:
+        1. Check if template references a direct upstream LLM node
+        2. If yes, only output the part AFTER that reference (suffix)
+        3. Prefix and LLM content have already been sent during LLM node streaming

-        示例：'{{start.test}}hahaha {{ llm_qa.output }} lalalalala a'
-        - 直接上游节点是 llm_qa
-        - 前缀 '{{start.test}}hahaha ' 已在 LLM 节点流式输出前发送
-        - LLM 内容在 LLM 节点流式输出
-        - End 节点只输出 ' lalalalala a'（后缀，一次性输出）
+        Note: Only LLM nodes get this special treatment. Other node types output normally.
+
+        Example: '{{start.test}}hahaha {{ llm_qa.output }} lalalalala a'
+        - Direct upstream LLM node is llm_qa
+        - Prefix '{{start.test}}hahaha ' was sent before LLM node streaming
+        - LLM content was streamed during LLM node execution
+        - End node only outputs ' lalalalala a' (suffix, sent as one chunk)

        Args:
-            state: 工作流状态
+            state: Workflow state

        Yields:
-            完成标记
+            Completion marker
        """
        logger.info(f"节点 {self.node_id} (End) 开始执行（流式）")

@@ -156,39 +159,45 @@ class EndNode(BaseNode):
            yield {"__final__": True, "result": output}
            return

-        # 找到直接上游节点
-        direct_upstream_nodes = []
+        # Find direct upstream LLM nodes
+        direct_upstream_llm_nodes = []
        for edge in self.workflow_config.get("edges", []):
            if edge.get("target") == self.node_id:
                source_node_id = edge.get("source")
-                direct_upstream_nodes.append(source_node_id)
+                # Check if the source node is an LLM node
+                for node in self.workflow_config.get("nodes", []):
+                    print("="*50)
+                    logger.info(f"节点 {self.node_id} 的类型 {node.get("type")}")
+                    if node.get("id") == source_node_id and node.get("type") == NodeType.LLM:
+                        direct_upstream_llm_nodes.append(source_node_id)
+                        break

-        logger.info(f"节点 {self.node_id} 的直接上游节点: {direct_upstream_nodes}")
+        logger.info(f"节点 {self.node_id} 的直接上游 LLM 节点: {direct_upstream_llm_nodes}")

-        # 解析模板部分
+        # Parse template parts
        parts = self._parse_template_parts(output_template, state)
        logger.info(f"节点 {self.node_id} 解析模板，共 {len(parts)} 个部分")
        for i, part in enumerate(parts):
            logger.info(f"[模板解析] part[{i}]: {part}")

-        # 找到第一个引用直接上游节点的动态引用
-        upstream_ref_index = None
+        # Find the first reference to a direct upstream LLM node
+        upstream_llm_ref_index = None
        for i, part in enumerate(parts):
-            if part["type"] == "dynamic" and part["node_id"] in direct_upstream_nodes:
-                upstream_ref_index = i
-                logger.info(f"节点 {self.node_id} 找到直接上游节点 {part['node_id']} 的引用，索引: {i}")
+            if part["type"] == "dynamic" and part["node_id"] in direct_upstream_llm_nodes:
+                upstream_llm_ref_index = i
+                logger.info(f"节点 {self.node_id} 找到直接上游 LLM 节点 {part['node_id']} 的引用，索引: {i}")
                break

-        if upstream_ref_index is None:
-            # 没有引用直接上游节点，输出完整模板内容
+        if upstream_llm_ref_index is None:
+            # No reference to direct upstream LLM node, output complete template content
            output = self._render_template(output_template, state)
-            logger.info(f"节点 {self.node_id} 没有引用直接上游节点，输出完整内容: '{output[:50]}...'")
+            logger.info(f"节点 {self.node_id} 没有引用直接上游 LLM 节点，输出完整内容: '{output[:50]}...'")

-            # 通过 writer 发送完整内容（作为一个 message chunk）
+            # Send complete content via writer (as a single message chunk)
            from langgraph.config import get_stream_writer
            writer = get_stream_writer()
            writer({
-                "type": "message",  # End 节点的输出使用 message 类型
+                "type": "message",  # End node output uses message type
                "node_id": self.node_id,
                "chunk": output,
                "full_content": output,
@@ -197,17 +206,17 @@ class EndNode(BaseNode):
            })
            logger.info(f"节点 {self.node_id} 已通过 writer 发送完整内容")

-            # yield 完成标记
+            # yield completion marker
            yield {"__final__": True, "result": output}
            return

-        # 有引用直接上游节点，只输出该引用之后的部分（后缀）
-        logger.info(f"节点 {self.node_id} 检测到直接上游节点引用，只输出后缀部分（从索引 {upstream_ref_index + 1} 开始）")
+        # Has reference to direct upstream LLM node, only output the part after that reference (suffix)
+        logger.info(f"节点 {self.node_id} 检测到直接上游 LLM 节点引用，只输出后缀部分（从索引 {upstream_llm_ref_index + 1} 开始）")

-        # 收集后缀部分
+        # Collect suffix parts
        suffix_parts = []
-        logger.info(f"[后缀调试] 开始收集后缀，从索引 {upstream_ref_index + 1} 到 {len(parts) - 1}")
-        for i in range(upstream_ref_index + 1, len(parts)):
+        logger.info(f"[后缀调试] 开始收集后缀，从索引 {upstream_llm_ref_index + 1} 到 {len(parts) - 1}")
+        for i in range(upstream_llm_ref_index + 1, len(parts)):
            part = parts[i]
            logger.info(f"[后缀调试] 处理 part[{i}]: {part}")
            if part["type"] == "static":
@@ -219,7 +228,7 @@ class EndNode(BaseNode):
                # Other dynamic references (if there are multiple references)
                node_id = part["node_id"]
                field = part["field"]
-                
+
                # Use VariablePool to get variable value
                pool = self.get_variable_pool(state)
                try:
@@ -232,7 +241,7 @@ class EndNode(BaseNode):

                # Convert to string if not None
                suffix_parts.append(str(content) if content is not None else "")
-        
+
        # 拼接后缀
        suffix = "".join(suffix_parts)

@@ -261,8 +270,8 @@ class EndNode(BaseNode):
            })
            logger.info(f"节点 {self.node_id} 已通过 writer 发送后缀，full_content 长度: {len(full_output)}")
        else:
-            logger.warning(f"[后缀调试] 节点 {self.node_id} 后缀为空，不发送！upstream_ref_index={upstream_ref_index}, parts数量={len(parts)}")
-        
+            logger.warning(f"[后缀调试] 节点 {self.node_id} 后缀为空，不发送！upstream_llm_ref_index={upstream_llm_ref_index}, parts数量={len(parts)}")
+
        # 统计信息
        node_outputs = state.get("node_outputs", {})
        total_nodes = len(node_outputs)
--- a/api/app/core/workflow/nodes/knowledge/init.py
+++ b/api/app/core/workflow/nodes/knowledge/init.py
@@ -0,0 +1,4 @@
+from app.core.workflow.nodes.knowledge.config import KnowledgeRetrievalNodeConfig
+from app.core.workflow.nodes.knowledge.node import KnowledgeRetrievalNode
+
+__all__ = ["KnowledgeRetrievalNode", "KnowledgeRetrievalNodeConfig"]
--- a/api/app/core/workflow/nodes/knowledge/config.py
+++ b/api/app/core/workflow/nodes/knowledge/config.py
@@ -0,0 +1,38 @@
+from uuid import UUID
+
+from pydantic import Field
+
+from app.core.workflow.nodes.base_config import BaseNodeConfig
+from app.schemas.chunk_schema import RetrieveType
+
+
+class KnowledgeRetrievalNodeConfig(BaseNodeConfig):
+    query: str = Field(
+        ...,
+        description="Search query string"
+    )
+
+    kb_ids: list[UUID] = Field(
+        ...,
+        description="Knowledge base IDs"
+    )
+
+    similarity_threshold: float = Field(
+        default=0.2,
+        description="Knowledge base similarity threshold"
+    )
+
+    vector_similarity_weight: float = Field(
+        default=0.3,
+        description="Knowledge base vector similarity weight"
+    )
+
+    top_k: int = Field(
+        default=4,
+        description="Knowledge base top k"
+    )
+
+    retrieve_type: RetrieveType = Field(
+        default=RetrieveType.PARTICIPLE,
+        description="Retrieve type"
+    )
--- a/api/app/core/workflow/nodes/knowledge/node.py
+++ b/api/app/core/workflow/nodes/knowledge/node.py
@@ -0,0 +1,93 @@
+import logging
+from typing import Any
+
+from app.core.rag.vdb.elasticsearch.elasticsearch_vector import ElasticSearchVectorFactory
+from app.core.workflow.nodes.base_node import BaseNode, WorkflowState
+from app.core.workflow.nodes.knowledge import KnowledgeRetrievalNodeConfig
+from app.db import get_db_context
+from app.models import knowledge_model, knowledgeshare_model
+from app.repositories import knowledge_repository
+from app.schemas.chunk_schema import RetrieveType
+from app.services import knowledge_service, knowledgeshare_service
+
+logger = logging.getLogger(__name__)
+
+
+class KnowledgeRetrievalNode(BaseNode):
+    def __init__(self, node_config: dict[str, Any], workflow_config: dict[str, Any]):
+        super().__init__(node_config, workflow_config)
+        self.typed_config = KnowledgeRetrievalNodeConfig(**self.config)
+
+    async def execute(self, state: WorkflowState) -> Any:
+        query = self._render_template(self.typed_config.query, state)
+        with get_db_context() as db:
+            filters = [
+                knowledge_model.Knowledge.id.in_(self.typed_config.kb_ids),
+                knowledge_model.Knowledge.permission_id == knowledge_model.PermissionType.Private,
+                knowledge_model.Knowledge.chunk_num > 0,
+                knowledge_model.Knowledge.status == 1
+            ]
+            existing_ids = knowledge_repository.get_chunked_knowledgeids(
+                db=db,
+                filters=filters
+            )
+            filters = [
+                knowledge_model.Knowledge.id.in_(self.typed_config.kb_ids),
+                knowledge_model.Knowledge.permission_id == knowledge_model.PermissionType.Share,
+                knowledge_model.Knowledge.chunk_num > 0,
+                knowledge_model.Knowledge.status == 1
+            ]
+            share_ids = knowledge_service.knowledge_repository.get_chunked_knowledgeids(
+                db=db,
+                filters=filters
+            )
+            if share_ids:
+                filters = [
+                    knowledgeshare_model.KnowledgeShare.target_kb_id.in_(self.typed_config.kb_ids)
+                ]
+                items = knowledgeshare_service.knowledgeshare_repository.get_source_kb_ids_by_target_kb_id(
+                    db=db,
+                    filters=filters
+                )
+                existing_ids.extend(items)
+
+            if not existing_ids:
+                raise RuntimeError("Knowledge base retrieval failed: the knowledge base does not exist.")
+
+            kb_id = existing_ids[0]
+            uuid_strs = [f"Vector_index_{kb_id}_Node".lower() for kb_id in existing_ids]
+            indices = ",".join(uuid_strs)
+
+            db_knowledge = knowledge_repository.get_knowledge_by_id(db=db, knowledge_id=kb_id)
+            if not db_knowledge:
+                raise RuntimeError("The knowledge base does not exist or access is denied.")
+
+            vector_service = ElasticSearchVectorFactory().init_vector(knowledge=db_knowledge)
+
+            match self.typed_config.retrieve_type:
+                case RetrieveType.PARTICIPLE:
+                    rs = vector_service.search_by_full_text(query=query, top_k=self.typed_config.top_k,
+                                                            indices=indices,
+                                                            score_threshold=self.typed_config.similarity_threshold)
+                    return [chunk.model_dump() for chunk in rs]
+                case RetrieveType.SEMANTIC:
+                    rs = vector_service.search_by_vector(query=query, top_k=self.typed_config.top_k,
+                                                         indices=indices,
+                                                         score_threshold=self.typed_config.vector_similarity_weight)
+                    return [chunk.model_dump() for chunk in rs]
+                case _:
+                    rs1 = vector_service.search_by_vector(query=query, top_k=self.typed_config.top_k,
+                                                          indices=indices,
+                                                          score_threshold=self.typed_config.vector_similarity_weight)
+                    rs2 = vector_service.search_by_full_text(query=query, top_k=self.typed_config.top_k,
+                                                             indices=indices,
+                                                             score_threshold=self.typed_config.similarity_threshold)
+                    # Efficient deduplication
+                    seen_ids = set()
+                    unique_rs = []
+                    for doc in rs1 + rs2:
+                        if doc.metadata["doc_id"] not in seen_ids:
+                            seen_ids.add(doc.metadata["doc_id"])
+                            unique_rs.append(doc)
+                    rs = vector_service.rerank(query=query, docs=unique_rs, top_k=self.typed_config.top_k)
+                    return [chunk.model_dump() for chunk in rs]
--- a/api/app/core/workflow/nodes/node_factory.py
+++ b/api/app/core/workflow/nodes/node_factory.py
@@ -7,7 +7,7 @@
 import logging
 from typing import Any, Union

-# from app.core.workflow.nodes.knowledge import KnowledgeRetrievalNode
+from app.core.workflow.nodes.knowledge import KnowledgeRetrievalNode
 from app.core.workflow.nodes.agent import AgentNode
 from app.core.workflow.nodes.base_node import BaseNode
 from app.core.workflow.nodes.end import EndNode
@@ -29,7 +29,7 @@ WorkflowNode = Union[
    AgentNode,
    TransformNode,
    AssignerNode,
-    # KnowledgeRetrievalNode,
+    KnowledgeRetrievalNode,
 ]


@@ -47,7 +47,7 @@ class NodeFactory:
        NodeType.AGENT: AgentNode,
        NodeType.TRANSFORM: TransformNode,
        NodeType.IF_ELSE: IfElseNode,
-        # NodeType.KNOWLEDGE_RETRIEVAL: KnowledgeRetrievalNode,
+        NodeType.KNOWLEDGE_RETRIEVAL: KnowledgeRetrievalNode,
        NodeType.ASSIGNER: AssignerNode,
    }

--- a/api/app/core/workflow/nodes/start/node.py
+++ b/api/app/core/workflow/nodes/start/node.py
@@ -23,7 +23,7 @@ class StartNode(BaseNode):
    
    注意：变量的验证和默认值处理由 Executor 在初始化时完成。
    """
-    
+
    def __init__(self, node_config: dict[str, Any], workflow_config: dict[str, Any]):
        """初始化 Start 节点
        
@@ -32,10 +32,10 @@ class StartNode(BaseNode):
            workflow_config: 工作流配置
        """
        super().__init__(node_config, workflow_config)
-        
+
        # 解析并验证配置
        self.typed_config = StartNodeConfig(**self.config)
-    
+
    async def execute(self, state: WorkflowState) -> dict[str, Any]:
        """执行 start 节点业务逻辑
        
@@ -48,13 +48,13 @@ class StartNode(BaseNode):
            包含系统参数、会话变量和自定义变量的字典
        """
        logger.info(f"节点 {self.node_id} (Start) 开始执行")
-        
+
        # 创建变量池实例（在方法内复用）
        pool = self.get_variable_pool(state)
-        
+
        # 处理自定义变量（传入 pool 避免重复创建）
        custom_vars = self._process_custom_variables(pool)
-        
+
        # 返回业务数据（包含自定义变量）
        result = {
            "message": pool.get("sys.message"),
@@ -64,14 +64,14 @@ class StartNode(BaseNode):
            "user_id": pool.get("sys.user_id"),
            **custom_vars  # 自定义变量作为节点输出的一部分
        }
-        
+
        logger.info(
            f"节点 {self.node_id} (Start) 执行完成，"
            f"输出了 {len(custom_vars)} 个自定义变量"
        )
-        
+
        return result
-    
+
    def _process_custom_variables(self, pool) -> dict[str, Any]:
        """处理自定义变量
        
@@ -88,34 +88,33 @@ class StartNode(BaseNode):
        """
        # 获取输入数据中的自定义变量
        input_variables = pool.get("sys.input_variables", default={})
-        
+
        processed = {}
-        
+
        # 遍历配置的变量定义
        for var_def in self.typed_config.variables:
            var_name = var_def.name
-            
+
            # 检查变量是否存在
            if var_name in input_variables:
                # 使用用户提供的值
                processed[var_name] = input_variables[var_name]
-            
+
            elif var_def.required:
                # 必需变量缺失
                raise ValueError(
                    f"缺少必需的输入变量: {var_name}"
                    + (f" ({var_def.description})" if var_def.description else "")
                )
-            
+
            elif var_def.default is not None:
                # 使用默认值
                processed[var_name] = var_def.default
                logger.debug(
                    f"变量 '{var_name}' 使用默认值: {var_def.default}"
                )
-        
+
        return processed
-    

    def _extract_input(self, state: WorkflowState) -> dict[str, Any]:
        """提取输入数据（用于记录）
@@ -127,7 +126,7 @@ class StartNode(BaseNode):
            输入数据字典
        """
        pool = self.get_variable_pool(state)
-        
+
        return {
            "execution_id": pool.get("sys.execution_id"),
            "conversation_id": pool.get("sys.conversation_id"),
--- a/api/app/repositories/knowledge_repository.py
+++ b/api/app/repositories/knowledge_repository.py
@@ -52,7 +52,7 @@ def get_knowledges_paginated(
        raise


-def get_chunded_knowledgeids(
+def get_chunked_knowledgeids(
        db: Session,
        filters: list
 ) -> list:
--- a/api/app/schemas/memory_storage_schema.py
+++ b/api/app/schemas/memory_storage_schema.py
@@ -31,7 +31,7 @@ class BaseDataSchema(BaseModel):
    # 保持原有必需字段为可选，以兼容不同数据源
    id: Optional[str] = Field(None, description="The unique identifier for the data entry.")
    statement: Optional[str] = Field(None, description="The statement text.")
-    created_at: str = Field(..., description="The creation timestamp in ISO 8601 format.")
+    created_at: Optional[str] = Field(None, description="The creation timestamp in ISO 8601 format.")
    expired_at: Optional[str] = Field(None, description="The expiration timestamp in ISO 8601 format.")
    description: Optional[str] = Field(None, description="The description of the data entry.")

@@ -46,6 +46,14 @@ class BaseDataSchema(BaseModel):
    relationship: Optional[Union[str, Dict[str, Any]]] = Field(None, description="The relationship object or string.")
    entity2: Optional[Dict[str, Any]] = Field(None, description="The second entity object.")

+    @model_validator(mode="before")
+    def _set_default_created_at(cls, v):
+        """Set default created_at if missing"""
+        if isinstance(v, dict) and v.get("created_at") is None:
+            from datetime import datetime
+            v["created_at"] = datetime.now().isoformat()
+        return v
+

 class QualityAssessmentSchema(BaseModel):
    """Schema for memory quality assessment results."""
--- a/api/app/services/knowledge_service.py
+++ b/api/app/services/knowledge_service.py
@@ -45,7 +45,7 @@ def get_chunded_knowledgeids(
    business_logger.debug(f"Query the list of vectorized knowledge base IDs: username={current_user.username}")

    try:
-        items = knowledge_repository.get_chunded_knowledgeids(
+        items = knowledge_repository.get_chunked_knowledgeids(
            db=db,
            filters=filters
        )
--- a/api/app/templates/workflows/simple_qa/template.yml
+++ b/api/app/templates/workflows/simple_qa/template.yml
@@ -44,11 +44,11 @@ nodes:
        - role: user
          content: "{{ sys.message }}"
      
-      model_id: gpt-3.5-turbo
+      model_id: null
      temperature: 0.7
      max_tokens: 1000
    position:
-      x: 300
+      x: 500
      y: 100

  - id: end
@@ -57,7 +57,7 @@ nodes:
    config:
      output: "{{ llm_qa.output }}"
    position:
-      x: 500
+      x: 900
      y: 100

 edges:
--- a/api/docker-compose.yml
+++ b/api/docker-compose.yml
@@ -3,7 +3,7 @@ version: '3.9'
 services:
  # MCP Server - standalone service
  mcp-server:
-    image: redbear-mem:latest
+    image: redbear-mem-open:latest
    container_name: mcp-server
    ports:
      - "8081:8081"  # MCP server port
@@ -28,14 +28,14 @@ services:

  # FastAPI application - connects to  MCP server
  api:
-    image: redbear-mem:latest
+    image: redbear-mem-open:latest
    container_name: api
    ports:
      - "8002:8000"
    env_file:
      - .env
    environment:
-      - MCP_SERVER_URL=http://mcp-server:8081
+      - MCP_SERVER_URL=http://mcp-server:8081  # Back to using container name
      - SERVER_IP=0.0.0.0  # Ensure MCP server binds to all interfaces
    volumes:
      - ./files:/files
@@ -51,12 +51,12 @@ services:

  # Celery worker - connects to MCP server
  worker:
-    image: redbear-mem:latest
+    image: redbear-mem-open:latest
    container_name: worker
    env_file:
      - .env
    environment:
-      - MCP_SERVER_URL=http://mcp-server:8081
+      - MCP_SERVER_URL=http://mcp-server:8081  # Back to using container name
    volumes:
      - ./files:/files
      - /etc/localtime:/etc/localtime:ro
--- a/api/env.example
+++ b/api/env.example
@@ -71,6 +71,18 @@ ENABLE_SINGLE_SESSION=
 MAX_FILE_SIZE=52428800  # 50MB:10 * 1024 * 1024
 FILE_PATH=/files

+# RAG Setting
+DOTNET_SYSTEM_GLOBALIZATION_INVARIANT=1
+HF_ENDPOINT=https://hf-mirror.com
+MINERU_EXECUTABLE=mineru
+MINERU_APISERVER=http://host.docker.internal:9987
+MINERU_OUTPUT_DIR=/files
+MINERU_BACKEND=pipeline
+MINERU_DELETE_OUTPUT=1
+TEXTLN_APISERVER=https://api.textin.com/ai/service/v1/pdf_to_markdown
+TEXTLN_APP_ID=
+TEXTLN_SECRET_CODE=
+
 # VOLC ASR
 VOLC_APP_KEY=
 VOLC_ACCESS_KEY=