Merge branch 'refs/heads/release/v0.3.1' into fix/Timebomb_031

api/app/celery_worker.py

@@ -2,6 +2,8 @@
 Celery Worker 入口点
 用于启动 Celery Worker: celery -A app.celery_worker worker --loglevel=info
 """
+from celery.signals import worker_process_init
+
 from app.celery_app import celery_app
 from app.core.logging_config import LoggingConfig, get_logger
 
@@ -13,4 +15,39 @@ logger.info("Celery worker logging initialized")
 # 导入任务模块以注册任务
 import app.tasks
 
+
+@worker_process_init.connect
+def _reinit_db_pool(**kwargs):
+    """
+    prefork 子进程启动时重建被 fork 污染的资源。
+    
+    fork() 后子进程继承了父进程的：
+    1. SQLAlchemy 连接池 — 多进程共享 TCP socket 导致 DB 连接损坏
+    2. ThreadPoolExecutor — fork 后线程状态不确定，第二个任务会死锁
+    """
+    # 重建 DB 连接池
+    from app.db import engine
+    engine.dispose()
+    logger.info("DB connection pool disposed for forked worker process")
+
+    # 重建模块级 ThreadPoolExecutor（fork 后线程池不可用）
+    try:
+        from app.core.rag.deepdoc.parser import figure_parser
+        from concurrent.futures import ThreadPoolExecutor
+        figure_parser.shared_executor = ThreadPoolExecutor(max_workers=10)
+        logger.info("figure_parser.shared_executor recreated")
+    except Exception as e:
+        logger.warning(f"Failed to recreate figure_parser.shared_executor: {e}")
+
+    try:
+        from app.core.rag.utils import libre_office
+        from concurrent.futures import ThreadPoolExecutor
+        import os
+        max_workers = os.cpu_count() * 2 if os.cpu_count() else 4
+        libre_office.executor = ThreadPoolExecutor(max_workers=max_workers)
+        logger.info("libre_office.executor recreated")
+    except Exception as e:
+        logger.warning(f"Failed to recreate libre_office.executor: {e}")
+
+
 __all__ = ['celery_app']

api/app/controllers/api_key_controller.py

@@ -167,6 +167,8 @@ def update_api_key(
 
         return success(data=api_key_schema.ApiKey.model_validate(api_key), msg="API Key 更新成功")
 
+    except BusinessException:
+        raise
     except Exception as e:
         logger.error(f"未知错误: {str(e)}", extra={
             "api_key_id": str(api_key_id),

api/app/controllers/app_controller.py

@@ -219,6 +219,7 @@ def delete_app(
 
 @router.post("/{app_id}/copy", summary="复制应用")
 @cur_workspace_access_guard()
+@check_app_quota
 def copy_app(
         app_id: uuid.UUID,
         new_name: Optional[str] = None,
@@ -1144,6 +1145,7 @@ async def import_workflow_config(
 
 @router.post("/workflow/import/save")
 @cur_workspace_access_guard()
+@check_app_quota
 async def save_workflow_import(
         data: WorkflowImportSave,
         db: Session = Depends(get_db),
@@ -1281,6 +1283,10 @@ async def import_app(
         return fail(msg="YAML 格式无效，缺少 app 字段", code=BizCode.BAD_REQUEST)
 
     target_app_id = uuid.UUID(app_id) if app_id else None
+    # 仅新建应用时检查配额，覆盖已有应用时跳过
+    if target_app_id is None:
+        from app.core.quota_manager import _check_quota
+        _check_quota(db, current_user.tenant_id, "app_quota", "app")
     result_app, warnings = AppDslService(db).import_dsl(
         dsl=dsl,
         workspace_id=current_user.current_workspace_id,

api/app/controllers/chunk_controller.py

@@ -457,7 +457,7 @@ async def retrieve_chunks(
                 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=retrieve_data.query, docs=unique_rs, top_k=retrieve_data.top_k)
+            rs = vector_service.rerank(query=retrieve_data.query, docs=unique_rs, top_k=retrieve_data.top_k) if unique_rs else []
             if retrieve_data.retrieve_type == chunk_schema.RetrieveType.Graph:
                 kb_ids = [str(kb_id) for kb_id in private_kb_ids]
                 workspace_ids = [str(workspace_id) for workspace_id in private_workspace_ids]

api/app/controllers/public_share_controller.py

@@ -219,9 +219,20 @@ def list_conversations(
     end_user_repo = EndUserRepository(db)
     app_service = AppService(db)
     app = app_service._get_app_or_404(share.app_id)
+    workspace_id = app.workspace_id
+
+    # 仅在新建终端用户时检查配额
+    existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id)
+    if existing_end_user is None:
+        from app.core.quota_manager import _check_quota
+        from app.models.workspace_model import Workspace
+        ws = db.query(Workspace).filter(Workspace.id == workspace_id).first()
+        if ws:
+            _check_quota(db, ws.tenant_id, "end_user_quota", "end_user")
+
     new_end_user = end_user_repo.get_or_create_end_user(
         app_id=share.app_id,
-        workspace_id=app.workspace_id,
+        workspace_id=workspace_id,
         other_id=other_id
     )
     logger.debug(new_end_user.id)
@@ -309,7 +320,6 @@ def get_conversation(
     "/chat",
     summary="发送消息（支持流式和非流式）"
 )
-@check_end_user_quota
 async def chat(
         payload: conversation_schema.ChatRequest,
         share_data: ShareTokenData = Depends(get_share_user_id),
@@ -350,6 +360,18 @@ async def chat(
         app_service = AppService(db)
         app = app_service._get_app_or_404(share.app_id)
         workspace_id = app.workspace_id
+
+        # 仅在新建终端用户时检查配额，已有用户复用不受限制
+        existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id)
+        logger.info(f"终端用户配额检查: workspace_id={workspace_id}, other_id={other_id}, existing={existing_end_user is not None}")
+        if existing_end_user is None:
+            from app.core.quota_manager import _check_quota
+            from app.models.workspace_model import Workspace
+            ws = db.query(Workspace).filter(Workspace.id == workspace_id).first()
+            if ws:
+                logger.info(f"新终端用户，执行配额检查: tenant_id={ws.tenant_id}")
+                _check_quota(db, ws.tenant_id, "end_user_quota", "end_user")
+
         new_end_user = end_user_repo.get_or_create_end_user(
             app_id=share.app_id,
             workspace_id=workspace_id,

api/app/controllers/service/app_api_controller.py

@@ -106,6 +106,16 @@ async def chat(
     other_id = payload.user_id
     workspace_id = api_key_auth.workspace_id
     end_user_repo = EndUserRepository(db)
+
+    # 仅在新建终端用户时检查配额，已有用户复用不受限制
+    existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id)
+    if existing_end_user is None:
+        from app.core.quota_manager import _check_quota
+        from app.models.workspace_model import Workspace
+        ws = db.query(Workspace).filter(Workspace.id == workspace_id).first()
+        if ws:
+            _check_quota(db, ws.tenant_id, "end_user_quota", "end_user")
+
     new_end_user = end_user_repo.get_or_create_end_user(
         app_id=app.id,
         workspace_id=workspace_id,

api/app/core/error_codes.py

@@ -32,6 +32,8 @@ class BizCode(IntEnum):
     API_KEY_DAILY_LIMIT_EXCEEDED = 3015
     API_KEY_QUOTA_EXCEEDED = 3016
     API_KEY_RATE_LIMIT_EXCEEDED = 3017
+    QUOTA_EXCEEDED = 3018
+    RATE_LIMIT_EXCEEDED = 3019
     # 资源（4xxx）
     NOT_FOUND = 4000
     USER_NOT_FOUND = 4001
@@ -156,6 +158,7 @@ HTTP_MAPPING = {
     BizCode.API_KEY_QPS_LIMIT_EXCEEDED: 429,
     BizCode.API_KEY_DAILY_LIMIT_EXCEEDED: 429,
     BizCode.API_KEY_QUOTA_EXCEEDED: 429,
+    BizCode.QUOTA_EXCEEDED: 402,
 
     BizCode.MODEL_CONFIG_INVALID: 400,
     BizCode.API_KEY_MISSING: 400,
@@ -185,4 +188,21 @@ HTTP_MAPPING = {
     BizCode.DB_ERROR: 500,
     BizCode.SERVICE_UNAVAILABLE: 503,
     BizCode.RATE_LIMITED: 429,
+    BizCode.RATE_LIMIT_EXCEEDED: 429,
+}
+
+ERROR_CODE_TO_BIZ_CODE = {
+    "QUOTA_EXCEEDED": BizCode.QUOTA_EXCEEDED,
+    "RATE_LIMIT_EXCEEDED": BizCode.RATE_LIMIT_EXCEEDED,
+    "API_KEY_NOT_FOUND": BizCode.API_KEY_NOT_FOUND,
+    "API_KEY_INVALID": BizCode.API_KEY_INVALID,
+    "API_KEY_EXPIRED": BizCode.API_KEY_EXPIRED,
+    "WORKSPACE_NOT_FOUND": BizCode.WORKSPACE_NOT_FOUND,
+    "WORKSPACE_NO_ACCESS": BizCode.WORKSPACE_NO_ACCESS,
+    "PERMISSION_DENIED": BizCode.PERMISSION_DENIED,
+    "TOKEN_EXPIRED": BizCode.TOKEN_EXPIRED,
+    "TOKEN_INVALID": BizCode.TOKEN_INVALID,
+    "VALIDATION_FAILED": BizCode.VALIDATION_FAILED,
+    "INVALID_PARAMETER": BizCode.INVALID_PARAMETER,
+    "MISSING_PARAMETER": BizCode.MISSING_PARAMETER,
 }

api/app/core/rag/common/connection_utils.py

@@ -33,18 +33,16 @@ def timeout(seconds: float | int | str = None, attempts: int = 2, *, exception:
             thread.daemon = True
             thread.start()
 
+            effective_timeout = seconds if seconds else 120  # 默认 120 秒超时
             for a in range(attempts):
                 try:
-                    if os.environ.get("ENABLE_TIMEOUT_ASSERTION"):
+                    result = result_queue.get(timeout=effective_timeout)
-                        result = result_queue.get(timeout=seconds)
-                    else:
-                        result = result_queue.get()
                     if isinstance(result, Exception):
                         raise result
                     return result
                 except queue.Empty:
                     pass
-            raise TimeoutError(f"Function '{func.__name__}' timed out after {seconds} seconds and {attempts} attempts.")
+            raise TimeoutError(f"Function '{func.__name__}' timed out after {effective_timeout} seconds and {attempts} attempts.")
 
         @wraps(func)
         async def async_wrapper(*args, **kwargs) -> Any:

api/app/core/rag/nlp/search.py

@@ -113,7 +113,7 @@ def knowledge_retrieval(
                 continue
 
         # Use the specified reranker for re-ranking
-        if reranker_id:
+        if reranker_id and all_results:
             try:
                 all_results = rerank(db=db, reranker_id=reranker_id, query=query, docs=all_results, top_k=reranker_top_k)
             except Exception as rerank_error:

api/app/core/rag/utils/es_conn.py

@@ -68,9 +68,9 @@ class ESConnection(DocStoreConnection):
         client_config = {
             "hosts": [hosts],
             "basic_auth": (os.getenv("ELASTICSEARCH_USERNAME", "elastic"), os.getenv("ELASTICSEARCH_PASSWORD", "elastic")),
-            "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 100000)),
+            "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 30)),
             "retry_on_timeout": os.getenv("ELASTICSEARCH_RETRY_ON_TIMEOUT", True) == "true",
-            "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 10000)),
+            "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 3)),
         }
 
         # Only add SSL settings if using HTTPS

api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py

@@ -1,25 +1,22 @@
 import os
 import logging
-from typing import Any, cast
+import threading
+from typing import Any
 from urllib.parse import urlparse
-import uuid
 
 import requests
 from elasticsearch import Elasticsearch, helpers
 from elasticsearch.helpers import BulkIndexError
 from packaging.version import parse as parse_version
-from pydantic import BaseModel, model_validator
-from abc import ABC
 # langchain-community
 # langchain-xinference
 # from langchain_community.embeddings import XinferenceEmbeddings
 # from langchain_xinference import XinferenceRerank
 from langchain_core.documents import Document
 from app.core.models.base import RedBearModelConfig
-from app.core.models import RedBearLLM, RedBearRerank
+from app.core.models import RedBearRerank
 from app.core.models.embedding import RedBearEmbeddings
-from app.models.models_model import ModelConfig, ModelApiKey
+from app.models.models_model import ModelApiKey
-from app.services.model_service import ModelConfigService
 
 from app.models.knowledge_model import Knowledge
 from app.core.rag.vdb.field import Field
@@ -29,37 +26,9 @@ from app.core.rag.models.chunk import DocumentChunk
 logger = logging.getLogger(__name__)
 
 
-class ElasticSearchConfig(BaseModel):
-    # Regular Elasticsearch config
-    host: str | None = None
-    port: int | None = None
-    username: str | None = None
-    password: str | None = None
-
-    # Common config
-    ca_certs: str | None = None
-    verify_certs: bool = False
-    request_timeout: int = 100000
-    retry_on_timeout: bool = True
-    max_retries: int = 10000
-
-    @model_validator(mode="before")
-    @classmethod
-    def validate_config(cls, values: dict):
-        # Regular Elasticsearch validation
-        if not values.get("host"):
-            raise ValueError("config HOST is required for regular Elasticsearch")
-        if not values.get("port"):
-            raise ValueError("config PORT is required for regular Elasticsearch")
-        if not values.get("username"):
-            raise ValueError("config USERNAME is required for regular Elasticsearch")
-        if not values.get("password"):
-            raise ValueError("config PASSWORD is required for regular Elasticsearch")
-        return values
-
-
 class ElasticSearchVector(BaseVector):
-    def __init__(self, index_name: str, config: ElasticSearchConfig, embedding_config: ModelApiKey, reranker_config: ModelApiKey):
+    def __init__(self, index_name: str, client: Elasticsearch,
+                 embedding_config: ModelApiKey, reranker_config: ModelApiKey):
         super().__init__(index_name.lower())
         
         # 初始化 Embedding 模型（自动支持火山引擎多模态）
@@ -77,58 +46,8 @@ class ElasticSearchVector(BaseVector):
             api_key=reranker_config.api_key,
             base_url=reranker_config.api_base
         ))
-        self._client = self._init_client(config)
+        # 使用外部传入的共享客户端
-        self._version = self._get_version()
+        self._client = client
-        self._check_version()
-
-    def _init_client(self, config: ElasticSearchConfig) -> Elasticsearch:
-        """
-        Initialize Elasticsearch client for regular Elasticsearch.
-        """
-        try:
-            # Regular Elasticsearch configuration
-            parsed_url = urlparse(config.host or "")
-            if parsed_url.scheme in {"http", "https"}:
-                hosts = f"{config.host}:{config.port}"
-                use_https = parsed_url.scheme == "https"
-            else:
-                hosts = f"https://{config.host}:{config.port}"
-                use_https = False
-
-            client_config = {
-                "hosts": [hosts],
-                "basic_auth": (config.username, config.password),
-                "request_timeout": config.request_timeout,
-                "retry_on_timeout": config.retry_on_timeout,
-                "max_retries": config.max_retries,
-            }
-
-            # Only add SSL settings if using HTTPS
-            if use_https:
-                client_config["verify_certs"] = config.verify_certs
-                if config.ca_certs:
-                    client_config["ca_certs"] = config.ca_certs
-
-            client = Elasticsearch(**client_config)
-
-            # Test connection
-            if not client.ping():
-                raise ConnectionError("Failed to connect to Elasticsearch")
-
-        except requests.ConnectionError as e:
-            raise ConnectionError(f"Vector database connection error: {str(e)}")
-        except Exception as e:
-            raise ConnectionError(f"Elasticsearch client initialization failed: {str(e)}")
-
-        return client
-
-    def _get_version(self) -> str:
-        info = self._client.info()
-        return cast(str, info["version"]["number"])
-
-    def _check_version(self):
-        if parse_version(self._version) < parse_version("8.0.0"):
-            raise ValueError("Elasticsearch vector database version must be greater than 8.0.0")
 
     def get_type(self) -> str:
         return "elasticsearch"
@@ -745,29 +664,79 @@ class ElasticSearchVector(BaseVector):
 
 
 class ElasticSearchVectorFactory:
-    @staticmethod
+    """ES 向量服务工厂 - 单例共享连接"""
-    def init_vector(knowledge: Knowledge) -> ElasticSearchVector:
+
+    _client: Elasticsearch | None = None
+    _lock = threading.Lock()
+    _version_checked = False
+
+    @classmethod
+    def _get_shared_client(cls) -> Elasticsearch:
+        """获取共享的 ES 客户端（线程安全的懒加载单例）"""
+        if cls._client is not None:
+            return cls._client
+
+        with cls._lock:
+            # 双重检查，防止并发时重复创建
+            if cls._client is not None:
+                return cls._client
+
+            try:
+                parsed_url = urlparse(os.getenv("ELASTICSEARCH_HOST", "127.0.0.1") or "")
+                if parsed_url.scheme in {"http", "https"}:
+                    hosts = f'{os.getenv("ELASTICSEARCH_HOST")}:{os.getenv("ELASTICSEARCH_PORT", 9200)}'
+                    use_https = parsed_url.scheme == "https"
+                else:
+                    hosts = f'https://{os.getenv("ELASTICSEARCH_HOST", "127.0.0.1")}:{os.getenv("ELASTICSEARCH_PORT", 9200)}'
+                    use_https = False
+
+                client_config = {
+                    "hosts": [hosts],
+                    "basic_auth": (
+                        os.getenv("ELASTICSEARCH_USERNAME", "elastic"),
+                        os.getenv("ELASTICSEARCH_PASSWORD", "elastic"),
+                    ),
+                    "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 30)),
+                    "retry_on_timeout": True,
+                    "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 3)),
+                    "connections_per_node": int(os.getenv("ELASTICSEARCH_CONNECTIONS_PER_NODE", 10)),
+                }
+
+                if use_https:
+                    client_config["verify_certs"] = os.getenv("ELASTICSEARCH_VERIFY_CERTS", "false") == "true"
+                    ca_certs = os.getenv("ELASTICSEARCH_CA_CERTS")
+                    if ca_certs:
+                        client_config["ca_certs"] = str(ca_certs)
+
+                client = Elasticsearch(**client_config)
+
+                if not client.ping():
+                    raise ConnectionError("Failed to connect to Elasticsearch")
+
+                # 版本检查只做一次
+                if not cls._version_checked:
+                    info = client.info()
+                    version = info["version"]["number"]
+                    if parse_version(version) < parse_version("8.0.0"):
+                        raise ValueError(f"Elasticsearch version must be >= 8.0.0, got {version}")
+                    cls._version_checked = True
+                    logger.info(f"Elasticsearch shared client initialized, version: {version}")
+
+                cls._client = client
+
+            except requests.ConnectionError as e:
+                raise ConnectionError(f"Vector database connection error: {str(e)}")
+            except Exception as e:
+                raise ConnectionError(f"Elasticsearch client initialization failed: {str(e)}")
+
+        return cls._client
+
+    @classmethod
+    def init_vector(cls, knowledge: Knowledge) -> ElasticSearchVector:
+        """创建向量服务实例（共享 ES 连接）"""
+        client = cls._get_shared_client()
         collection_name = f"Vector_index_{knowledge.id}_Node"
 
-        # Use regular Elasticsearch with config values
-        config_dict = {
-            "host": os.getenv("ELASTICSEARCH_HOST", "127.0.0.1"),
-            "port": os.getenv("ELASTICSEARCH_PORT", 9200),
-            "username": os.getenv("ELASTICSEARCH_USERNAME", "elastic"),
-            "password": os.getenv("ELASTICSEARCH_PASSWORD", "elastic"),
-        }
-
-        # Common configuration
-        config_dict.update(
-            {
-                "ca_certs": str(os.getenv("ELASTICSEARCH_CA_CERTS")) if os.getenv("ELASTICSEARCH_CA_CERTS") else None,
-                "verify_certs": os.getenv("ELASTICSEARCH_VERIFY_CERTS", False) == "true",
-                "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 100000)),
-                "retry_on_timeout": os.getenv("ELASTICSEARCH_RETRY_ON_TIMEOUT", True) == "true",
-                "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 10000)),
-            }
-        )
-
         if knowledge.embedding is None:
             raise ValueError(f"embedding_id config error: {str(knowledge.embedding_id)}")
         if knowledge.reranker is None:
@@ -775,9 +744,9 @@ class ElasticSearchVectorFactory:
 
         return ElasticSearchVector(
             index_name=collection_name,
-            config=ElasticSearchConfig(**config_dict),
+            client=client,
             embedding_config=knowledge.embedding.api_keys[0],
-            reranker_config=knowledge.reranker.api_keys[0]
+            reranker_config=knowledge.reranker.api_keys[0],
         )
 
 

api/app/i18n/exceptions.py

@@ -6,12 +6,14 @@ error messages based on the current request's language.
 """
 
 import logging
+import time
 from contextvars import ContextVar
 from typing import Any, Dict, Optional
 
 from fastapi import HTTPException, Request
 
 from app.i18n.service import get_translation_service
+from app.core.error_codes import ERROR_CODE_TO_BIZ_CODE, BizCode
 
 logger = logging.getLogger(__name__)
 
@@ -118,15 +120,24 @@ class I18nException(HTTPException):
             **params
         )
 
-        # Build error detail
+        # Convert error_code string to BizCode value
-        detail = {
+        biz_code = ERROR_CODE_TO_BIZ_CODE.get(
-            "error_code": self.error_code,
+            self.error_code,
-            "message": message,
+            BizCode.BAD_REQUEST
-        }
+        )
 
-        # Add parameters to detail if provided
+        # Build error detail in standard format for compatibility
-        if params:
+        # main.py handler expects "message" and "error_code" fields for filtering
-            detail["params"] = params
+        # but we also include standard format fields
+        detail = {
+            "code": biz_code.value,
+            "msg": message,
+            "message": message,
+            "error_code": self.error_code,
+            "data": params if params else {},
+            "error": message,
+            "time": int(time.time() * 1000),
+        }
 
         # Initialize HTTPException
         super().__init__(

api/app/repositories/end_user_repository.py

@@ -66,6 +66,17 @@ class EndUserRepository:
             db_logger.error(f"查询宿主 {end_user_id} 时出错: {str(e)}")
             raise
 
+    def get_end_user_by_other_id(self, workspace_id: uuid.UUID, other_id: str) -> Optional["EndUser"]:
+        """按 workspace_id + other_id 查找终端用户，不存在返回 None"""
+        return (
+            self.db.query(EndUser)
+            .filter(
+                EndUser.workspace_id == workspace_id,
+                EndUser.other_id == other_id
+            )
+            .first()
+        )
+
     def get_or_create_end_user(
         self,
         app_id: uuid.UUID,

api/app/services/api_key_service.py

@@ -51,7 +51,7 @@ class ApiKeyService:
             if existing:
                 raise BusinessException(f"API Key 名称 {data.name} 已存在", BizCode.API_KEY_DUPLICATE_NAME)
 
-            # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，自动截断到套餐上限
+            # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，直接报错
             from app.models.workspace_model import Workspace
             from app.core.quota_manager import get_api_ops_rate_limit
 
@@ -59,7 +59,10 @@ class ApiKeyService:
             if workspace:
                 tenant_api_ops_limit = get_api_ops_rate_limit(db, workspace.tenant_id)
                 if tenant_api_ops_limit and data.rate_limit > tenant_api_ops_limit:
-                    data.rate_limit = tenant_api_ops_limit
+                    raise BusinessException(
+                        f"API Key QPS 不能超过套餐上限 {tenant_api_ops_limit}",
+                        BizCode.BAD_REQUEST
+                    )
 
             # 生成 API Key
             api_key = generate_api_key(data.type)
@@ -162,7 +165,7 @@ class ApiKeyService:
             if existing:
                 raise BusinessException(f"API Key 名称 {data.name} 已存在", BizCode.API_KEY_DUPLICATE_NAME)
 
-        # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，自动截断到套餐上限
+        # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，直接报错
         if data.rate_limit is not None:
             from app.models.workspace_model import Workspace
             from app.core.quota_manager import get_api_ops_rate_limit
@@ -171,7 +174,10 @@ class ApiKeyService:
             if workspace:
                 tenant_api_ops_limit = get_api_ops_rate_limit(db, workspace.tenant_id)
                 if tenant_api_ops_limit and data.rate_limit > tenant_api_ops_limit:
-                    data.rate_limit = tenant_api_ops_limit
+                    raise BusinessException(
+                        f"API Key QPS 不能超过套餐上限 {tenant_api_ops_limit}",
+                        BizCode.BAD_REQUEST
+                    )
 
         update_data = data.model_dump(exclude_unset=True)
         ApiKeyRepository.update(db, api_key_id, update_data)

api/app/services/app_dsl_service.py

@@ -434,19 +434,37 @@ class AppDslService:
     def _resolve_model(self, ref: Optional[dict], tenant_id: uuid.UUID, warnings: list) -> Optional[uuid.UUID]:
         if not ref:
             return None
-        q = self.db.query(ModelConfig).filter(
+        model_id = ref.get("id")
-            ModelConfig.tenant_id == tenant_id,
+        if model_id:
-            ModelConfig.name == ref.get("name"),
+            try:
-            ModelConfig.is_active.is_(True)
+                model_uuid = uuid.UUID(str(model_id))
-        )
+                m = self.db.query(ModelConfig).filter(
-        if ref.get("provider"):
+                    ModelConfig.id == model_uuid,
-            q = q.filter(ModelConfig.provider == ref["provider"])
+                    ModelConfig.tenant_id == tenant_id,
-        if ref.get("type"):
+                    ModelConfig.is_active.is_(True)
-            q = q.filter(ModelConfig.type == ref["type"])
+                ).first()
-        m = q.first()
+                if m:
-        if not m:
+                    return str(m.id)
-            warnings.append(f"模型 '{ref.get('name')}' 未匹配，已置空，请导入后手动配置")
+            except (ValueError, AttributeError):
-        return m.id if m else None
+                pass
+        model_name = ref.get("name")
+        if model_name:
+            q = self.db.query(ModelConfig).filter(
+                ModelConfig.tenant_id == tenant_id,
+                ModelConfig.name == model_name,
+                ModelConfig.is_active.is_(True)
+            )
+            if ref.get("provider"):
+                q = q.filter(ModelConfig.provider == ref["provider"])
+            if ref.get("type"):
+                q = q.filter(ModelConfig.type == ref["type"])
+            m = q.first()
+            if m:
+                return str(m.id)
+            warnings.append(f"模型 '{model_name}' 未匹配，已置空，请导入后手动配置")
+        else:
+            warnings.append(f"模型 ID '{model_id}' 未匹配，已置空，请导入后手动配置")
+        return None
 
     def _resolve_kb(self, ref: Optional[dict], workspace_id: uuid.UUID, warnings: list) -> Optional[str]:
         if not ref:
@@ -587,7 +605,7 @@ class AppDslService:
                     if not kb_id:
                         continue
                     kb_ref = {}
-                    if isinstance(kb_id, str) and len(kb_id) >= 36:
+                    if isinstance(kb_id, str):
                         try:
                             uuid.UUID(kb_id)
                             kb_ref["id"] = kb_id
@@ -601,6 +619,33 @@ class AppDslService:
                     else:
                         warnings.append(f"[{node_label}] 知识库 '{kb_id}' 未匹配，已移除，请导入后手动配置")
                 config["knowledge_bases"] = resolved_kbs
+            elif node_type in (NodeType.LLM.value, NodeType.QUESTION_CLASSIFIER.value, NodeType.PARAMETER_EXTRACTOR.value):
+                model_ref = config.get("model_id")
+                if model_ref:
+                    ref_dict = None
+                    if isinstance(model_ref, dict):
+                        ref_id = model_ref.get("id")
+                        ref_name = model_ref.get("name")
+                        if ref_id:
+                            ref_dict = {"id": ref_id}
+                        elif ref_name is not None:
+                            ref_dict = {"name": ref_name, "provider": model_ref.get("provider"), "type": model_ref.get("type")}
+                    elif isinstance(model_ref, str):
+                        try:
+                            uuid.UUID(model_ref)
+                            ref_dict = {"id": model_ref}
+                        except ValueError:
+                            ref_dict = {"name": model_ref}
+                    if ref_dict:
+                        resolved_model_id = self._resolve_model(ref_dict, tenant_id, warnings)
+                        if resolved_model_id:
+                            config["model_id"] = resolved_model_id
+                        else:
+                            warnings.append(f"[{node_label}] 模型未匹配，已置空，请导入后手动配置")
+                            config["model_id"] = None
+                    else:
+                        warnings.append(f"[{node_label}] 模型未匹配，已置空，请导入后手动配置")
+                        config["model_id"] = None
             resolved_nodes.append({**node, "config": config})
         return resolved_nodes
 

api/app/services/knowledge_service.py

@@ -7,7 +7,6 @@ from app.models.models_model import ModelConfig
 from app.schemas.knowledge_schema import KnowledgeCreate, KnowledgeUpdate
 from app.repositories import knowledge_repository
 from app.core.logging_config import get_business_logger
-from app.repositories.model_repository import ModelConfigRepository
 from app.models.models_model import ModelType
 
 business_logger = get_business_logger()
@@ -77,53 +76,32 @@ def create_knowledge(
 
         tenant_id = workspace.tenant_id
 
-        def _get_model_by_name_or_fallback(model_name: str | None, model_types: list, label: str):
-            """优先按 workspace 指定的 model name 查，找不到再 fallback 到 tenant 下第一个"""
-            if model_name:
-                model = db.query(ModelConfig).filter(
-                    ModelConfig.tenant_id == tenant_id,
-                    ModelConfig.name == model_name,
-                    ModelConfig.type.in_([t.value for t in model_types]),
-                    ModelConfig.is_active == True,
-                    ModelConfig.is_composite == False
-                ).first()
-                if model:
-                    business_logger.debug(f"Auto-bind {label} model from workspace default: {model.id} ({model_name})")
-                    return model
-                business_logger.debug(f"Workspace default {label} model '{model_name}' not found, falling back to tenant")
-            models = ModelConfigRepository.get_by_type(db=db, model_types=model_types, tenant_id=tenant_id, is_active=True)
-            if models:
-                business_logger.debug(f"Auto-bind {label} model from tenant fallback: {models[0].id}")
-                return models[0]
-            return None
-
         if not knowledge.embedding_id:
-            model = _get_model_by_name_or_fallback(workspace.embedding, [ModelType.EMBEDDING], "embedding")
+            if not workspace.embedding:
-            if model:
+                raise Exception("工作空间未配置 Embedding 模型，请先完善工作空间配置后重试")
-                knowledge.embedding_id = model.id
+            knowledge.embedding_id = workspace.embedding
 
         if not knowledge.reranker_id:
-            model = _get_model_by_name_or_fallback(workspace.rerank, [ModelType.RERANK], "rerank")
+            if not workspace.rerank:
-            if model:
+                raise Exception("工作空间未配置 Rerank 模型，请先完善工作空间配置后重试")
-                knowledge.reranker_id = model.id
+            knowledge.reranker_id = workspace.rerank
 
         if not knowledge.llm_id:
-            model = _get_model_by_name_or_fallback(workspace.llm, [ModelType.LLM, ModelType.CHAT], "llm")
+            if not workspace.llm:
-            if model:
+                raise Exception("工作空间未配置 LLM 模型，请先完善工作空间配置后重试")
-                knowledge.llm_id = model.id
+            knowledge.llm_id = workspace.llm
 
         if not knowledge.image2text_id:
-            image2text_models = db.query(ModelConfig).filter(
+            model = db.query(ModelConfig).filter(
                 ModelConfig.tenant_id == tenant_id,
-                ModelConfig.type.in_([ModelType.CHAT.value]),
+                ModelConfig.type.in_([ModelType.CHAT.value, ModelType.LLM.value]),
                 ModelConfig.capability.contains(["vision"]),
                 ModelConfig.is_active == True,
-                ModelConfig.is_composite == False
+            ).order_by(ModelConfig.created_at.desc()).first()
-            ).order_by(ModelConfig.created_at.desc()).all()
+            if not model:
-            if not image2text_models:
                 raise Exception("租户下没有可用的视觉模型，创建知识库失败")
-            knowledge.image2text_id = image2text_models[0].id
+            knowledge.image2text_id = model.id
-            business_logger.debug(f"Auto-bind image2text model: {image2text_models[0].id}")
+            business_logger.debug(f"Auto-bind image2text model: {model.id}")
 
         business_logger.debug(f"Start creating the knowledge base: {knowledge.name}")
         db_knowledge = knowledge_repository.create_knowledge(

api/app/tasks.py

@@ -251,8 +251,40 @@ def parse_document(file_path: str, document_id: uuid.UUID):
         # Prepare vision_model for parsing
         vision_model = _build_vision_model(file_path, db_knowledge)
 
+        # 先将文件读入内存，避免解析过程中依赖 NFS 文件持续可访问
+        # python-docx 等库在 binary=None 时会用路径直接打开文件，
+        # 在 NFS/共享存储上可能因缓存失效导致 "Package not found"
+        max_wait_seconds = 30
+        wait_interval = 2
+        waited = 0
+        file_binary = None
+        while waited <= max_wait_seconds:
+            # os.listdir 强制 NFS 客户端刷新目录缓存
+            parent_dir = os.path.dirname(file_path)
+            try:
+                os.listdir(parent_dir)
+            except OSError:
+                pass
+            try:
+                with open(file_path, "rb") as f:
+                    file_binary = f.read()
+                if not file_binary:
+                    # NFS 上文件存在但内容为空（可能还在同步中）
+                    raise IOError(f"File is empty (0 bytes), NFS may still be syncing: {file_path}")
+                break
+            except (FileNotFoundError, IOError) as e:
+                if waited >= max_wait_seconds:
+                    raise type(e)(
+                        f"File not accessible at '{file_path}' after waiting {max_wait_seconds}s: {e}"
+                    )
+                logger.warning(f"File not ready on this node, retrying in {wait_interval}s: {file_path} ({e})")
+                time.sleep(wait_interval)
+                waited += wait_interval
+
         from app.core.rag.app.naive import chunk
+        logger.info(f"[ParseDoc] file_binary size={len(file_binary)} bytes, type={type(file_binary).__name__}, bool={bool(file_binary)}")
         res = chunk(filename=file_path,
+                    binary=file_binary,
                     from_page=0,
                     to_page=DEFAULT_PARSE_TO_PAGE,
                     callback=progress_callback,

web/src/components/SiderMenu/SubscriptionDetailModal.tsx

@@ -2,11 +2,11 @@
  * @Author: ZhaoYing 
  * @Date: 2026-04-14 12:28:23 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-04-16 17:34:02
+ * @Last Modified time: 2026-04-21 15:46:35
  */
 
 import { useState, forwardRef, useImperativeHandle } from 'react';
-import { Flex, Tooltip, Divider } from 'antd';
+import { Flex, Divider } from 'antd';
 import { useTranslation } from 'react-i18next';
 import clsx from 'clsx';
 
@@ -82,8 +82,7 @@ const SubscriptionDetailModal = forwardRef<SubscriptionDetailModalRef>((_props,
       {/* Features */}
       <Flex gap={12} vertical className="rb:space-y-3 rb:mb-4 rb:h-[calc(100vh-341px)]! rb:overflow-y-auto">
         {billingUnits.map(({ key, unit, icon }) => {
-          const value = detail?.quotas[key as keyof Subscription['quotas']];
+          const value = detail?.quotas?.[key as keyof Subscription['quotas']];
-          if (value === undefined || value === null) return null;
           return (
             <UnitWrapper
               key={key}
@@ -95,7 +94,7 @@ const SubscriptionDetailModal = forwardRef<SubscriptionDetailModalRef>((_props,
             />
           )
         })}
-        {detail?.package_plan?.tech_support && (
+        {detail?.package_plan?.tech_support && detail?.package_plan?.[getKeyWithLanguage('tech_support')] && (
           <UnitWrapper
             titleKey="tech_support"
             value={String(detail?.package_plan?.[getKeyWithLanguage('tech_support')] ?? '')}
@@ -103,7 +102,7 @@ const SubscriptionDetailModal = forwardRef<SubscriptionDetailModalRef>((_props,
             theme_color={detail?.package_plan?.theme_color}
           />
         )}
-        {detail?.package_plan?.sla_compliance && (
+        {detail?.package_plan?.sla_compliance && detail?.package_plan?.[getKeyWithLanguage('sla_compliance')] && (
           <UnitWrapper
             titleKey="sla"
             value={String(detail?.package_plan?.[getKeyWithLanguage('sla_compliance')] ?? '')}

web/src/components/SiderMenu/index.tsx

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-02-02 15:25:31 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-04-20 10:15:20 
+ * @Last Modified time: 2026-04-21 15:46:03
  */
 /**
  * SiderMenu Component
@@ -417,7 +417,7 @@ const Menu: FC<{
           <div className="rb:grid rb:grid-cols-4 rb:mt-4">
             {['workspace_quota', 'skill_quota', 'app_quota', 'model_quota'].map(key => (
               <div key={key} className="rb:text-center">
-                <div className="rb:text-[13px] rb:font-[MiSans-Semibold] rb:font-semibold">{subscription.quotas?.[key as keyof typeof subscription.quotas]}</div>
+                <div className="rb:text-[13px] rb:font-[MiSans-Semibold] rb:font-semibold">{subscription.quotas?.[key as keyof typeof subscription.quotas] ?? t('package.noLimit')}</div>
                 <div className="rb:mt-1 rb:text-[#5B6167] rb:text-[10px] rb:leading-3.5">{t(`index.${key}`)}</div>
               </div>
             ))}

web/src/i18n/en.ts

@@ -451,6 +451,9 @@ export const en = {
       logoutApiCannotRefreshToken: 'Logout API cannot refresh token',
       publicApiCannotRefreshToken: 'Public API cannot refresh token',
       refreshTokenNotExist: 'Refresh token does not exist',
+      SYSTEM_DEFAULT_SCENE_CANNOT_DELETE: 'This is a system preset scene and cannot be deleted',
+      SYSTEM_DEFAULT_CLASS_CANNOT_DELETE: 'This scene is a system preset scene and cannot be deleted',
+      SYSTEM_DEFAULT_SCENE_CANNOT_UPDATE: 'This scene is a system preset scene and cannot be modified',
       reset: 'Reset',
       refresh: 'Refresh',
       return: 'Return',
@@ -3093,6 +3096,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re
       editPackage: 'Edit Package',
 
       viewDetail: 'View full package details',
+      noLimit: 'Unlimited',
     },
   },
 };

web/src/i18n/zh.ts

@@ -1130,6 +1130,9 @@ export const zh = {
       logoutApiCannotRefreshToken: '退出登录接口不能刷新token',
       publicApiCannotRefreshToken: '公共接口不能刷新token',
       refreshTokenNotExist: '刷新token不存在',
+      SYSTEM_DEFAULT_SCENE_CANNOT_DELETE: '该场景为系统预设场景，不允许删除',
+      SYSTEM_DEFAULT_CLASS_CANNOT_DELETE: '该场景为系统预设场景，不允许删除',
+      SYSTEM_DEFAULT_SCENE_CANNOT_UPDATE: '该场景为系统预设场景，不允许修改',
       reset: '重置',
       refresh: '刷新',
       return: '返回',
@@ -3057,6 +3060,7 @@ export const zh = {
       editPackage: '编辑套餐',
 
       viewDetail: '查看完整套餐详情',
+      noLimit: '不限制',
     },
   },
 }

web/src/utils/stream.ts

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-02-02 16:35:43 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-03-18 14:32:40
+ * @Last Modified time: 2026-04-22 10:16:43
  */
 /**
  * Server-Sent Events (SSE) Stream Utility Module
@@ -176,12 +176,12 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe
       case 500:
       case 502:
         const errorData = await response.json();
-        const errorInfo = errorData.error || i18n.t('common.serviceUpgrading');
+        const errorInfo = errorData.error || errorData.msg || i18n.t('common.serviceUpgrading');
         message.warning(errorInfo);
         throw new Error(errorData);
       case 400:
         const error = await response.json();
-        const error400 = error.error || 'Bad Request';
+        const error400 = error.error || error.msg || 'Bad Request';
         message.warning(error400);
         throw new Error(error);
       case 403:
@@ -190,7 +190,7 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe
         throw new Error(errors);
       case 504:
         const errorJson = await response.json();
-        const errorMsg = errorJson.error || i18n.t('common.serverError');
+        const errorMsg = errorJson.error || errorJson.msg || i18n.t('common.serverError');
         message.warning(errorMsg);
         throw new Error(errorJson);
       case 401:
@@ -204,6 +204,13 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe
           return;
         }
         break;
+      default:
+        if (!response.ok) {
+          const defaultData = await response.json().catch(() => ({}));
+          const defaultMsg = defaultData.error || defaultData.msg;
+          if (defaultMsg) message.warning(defaultMsg);
+          throw new Error(defaultMsg || `HTTP ${response.status}`);
+        }
     }
     if (!response.body) throw new Error('No response body');
 

web/src/views/ApplicationConfig/components/Knowledge/Knowledge.tsx

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-02-03 16:25:32 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-03-04 10:34:43
+ * @Last Modified time: 2026-04-21 13:34:52
  */
 /**
  * Knowledge Base Component
@@ -54,7 +54,7 @@ const Knowledge: FC<{value?: KnowledgeConfig; onChange?: (config: KnowledgeConfi
       const basesWithoutName = knowledge_bases.filter(base => !base.name)
       if (basesWithoutName.length > 0) {
         // Call API to get complete knowledge base information
-        getKnowledgeBaseList().then(res => {
+        getKnowledgeBaseList(undefined, { kb_ids: basesWithoutName.map(vo => vo.kb_id).join(',') }).then(res => {
           const fullBases = knowledge_bases.map(base => {
             if (!base.name) {
               const fullBase = res.items.find((item: any) => item.id === base.kb_id)

web/src/views/ModelManagement/components/CustomModelModal.tsx

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-02-03 16:49:28 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-04-16 18:03:53
+ * @Last Modified time: 2026-04-21 15:02:53
  */
 /**
  * Custom Model Modal
@@ -230,21 +230,23 @@ const CustomModelModal = forwardRef<CustomModelModalRef, CustomModelModalProps>(
           <Input.TextArea placeholder={t('common.pleaseEnter')} />
         </Form.Item>
 
-        <Form.Item
+        {!isEdit && <>
-          name={["api_keys", 0, "api_key"]}
+          <Form.Item
-          label={t('modelNew.api_key')}
+            name={["api_keys", 0, "api_key"]}
-          rules={[{ required: true, message: t('common.inputPlaceholder', { title: t('modelNew.api_key') }) }]}
+            label={t('modelNew.api_key')}
-        >
+            rules={[{ required: true, message: t('common.inputPlaceholder', { title: t('modelNew.api_key') }) }]}
-          <Input.Password placeholder={t('common.pleaseEnter')} />
+          >
-        </Form.Item>
+            <Input.Password placeholder={t('common.pleaseEnter')} />
+          </Form.Item>
 
-        <Form.Item
+          <Form.Item
-          name={["api_keys", 0, "api_base"]}
+            name={["api_keys", 0, "api_base"]}
-          label={t('modelNew.api_base')}
+            label={t('modelNew.api_base')}
-          rules={[{ required: true, message: t('common.inputPlaceholder', { title: t('modelNew.api_base') }) }]}
+            rules={[{ required: true, message: t('common.inputPlaceholder', { title: t('modelNew.api_base') }) }]}
-        >
+          >
-          <Input placeholder="https://api.example.com/v1" />
+            <Input placeholder="https://api.example.com/v1" />
-        </Form.Item>
+          </Form.Item>
+        </>}
 
         {['llm', 'chat'].includes(modelType as string) &&
           <Row gutter={16}>

web/src/views/Package/constant.ts

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-04-14 11:43:57 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-04-14 11:44:40
+ * @Last Modified time: 2026-04-21 15:44:13
  */
 export const billingUnits = [
   {
@@ -42,7 +42,7 @@ export const billingUnits = [
   },
   {
     key: 'model_quota',
-    unit: 'ops', placeholder: 'numberPlaceholder',
+    unit: 'pcs', placeholder: 'numberPlaceholder',
     icon: 'model',
   },
   {

web/src/views/Package/index.tsx

@@ -2,7 +2,7 @@
  * @Author: ZhaoYing 
  * @Date: 2026-04-14 11:34:42 
  * @Last Modified by: ZhaoYing
- * @Last Modified time: 2026-04-16 17:23:49
+ * @Last Modified time: 2026-04-21 15:45:30
  */
 /**
  * Package Component
@@ -60,7 +60,7 @@ const btnClassNames = {
   default: 'rb:h-10! rb:rounded-[8px]! rb:bg-[#212332]! rb:text-white! rb:border-0! rb:hover:border-0! rb:hover:opacity-[0.8]',
 }
 
-export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#171719' }: { titleKey: string; value: number | string; icon: string; unit?: string; theme_color?: string; }) => {
+export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#171719' }: { titleKey: string; value?: number | string | null; icon: string; unit?: string; theme_color?: string; }) => {
   const { t } = useTranslation();
 
   const renderFeatureIcon = (iconKey: string, color: string) => {
@@ -78,7 +78,7 @@ export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#17171
       >{renderFeatureIcon(icon, theme_color)}</Flex>
       <div className="rb:text-[13px] rb:leading-4.5">
         <div className="rb:text-[#5F6266]">{t(`package.${titleKey}`)}</div>
-        <div>{value} {unit ? t(`package.${unit}`) : ''}</div>
+        {value ? <div>{value} {unit ? t(`package.${unit}`) : ''}</div> : <div>{t('package.noLimit')}</div>}
       </div>
     </Flex>
   )
@@ -252,7 +252,6 @@ const Package: FC = () => {
                     >
                       {billingUnits.map(({ key, unit, icon }) => {
                         const value = pkg?.quotas?.[key as keyof Package['quotas']];
-                        if (value === undefined || value === null) return null;
                         return (
                           <UnitWrapper
                             key={key}
@@ -264,7 +263,7 @@ const Package: FC = () => {
                           />
                         )
                       })}
-                      {pkg.tech_support && (
+                      {pkg.tech_support && pkg[getKeyWithLanguage('tech_support')] && (
                         <UnitWrapper
                             titleKey="tech_support"
                             value={String(pkg[getKeyWithLanguage('tech_support')] ?? '')}
@@ -272,7 +271,7 @@ const Package: FC = () => {
                             theme_color={pkg.theme_color}
                           />
                       )}
-                      {pkg.sla_compliance && (
+                      {pkg.sla_compliance && pkg[getKeyWithLanguage('sla_compliance')] && (
                         <UnitWrapper
                           titleKey="sla"
                           value={String(pkg[getKeyWithLanguage('sla_compliance')] ?? '')}

web/src/views/Workflow/components/Properties/Knowledge/Knowledge.tsx

@@ -29,12 +29,13 @@ const Knowledge: FC<{value?: KnowledgeConfig; onChange?: (config: KnowledgeConfi
     if (value && JSON.stringify(value) !== JSON.stringify(editConfig)) {
       setEditConfig({ ...(value || {}) })
       const knowledge_bases = [...(value.knowledge_bases || [])]
+      setKnowledgeList(knowledge_bases)
       
       // 检查是否有knowledge_bases缺少name字段
       const basesWithoutName = knowledge_bases.filter(base => !base.name)
       if (basesWithoutName.length > 0) {
         // 调用接口获取完整的知识库信息
-        getKnowledgeBaseList().then(res => {
+        getKnowledgeBaseList(undefined, { kb_ids: basesWithoutName.map(vo => vo.kb_id).join(',') }).then(res => {
           const fullBases = knowledge_bases.map(base => {
             if (!base.name) {
               const fullBase = res.items.find((item: any) => item.id === base.kb_id)

web/src/views/Workflow/components/Properties/ToolConfig/index.tsx

@@ -1,4 +1,4 @@
-import { type FC, useEffect, useState, useMemo } from "react";
+import { type FC, useEffect, useState } from "react";
 import { useTranslation } from 'react-i18next'
 import { Form, Select, Switch, Cascader, type CascaderProps, Tooltip } from 'antd'
 import type { Suggestion } from '../../Editor/plugin/AutocompletePlugin'
@@ -155,18 +155,18 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({
     const targetOption = selectedOptions[selectedOptions.length - 1];
     const curParameters = [...(targetOption.parameters ?? [])]
     setParameters([...curParameters])
-    const inititalValue: any = { tool_id: selectedOptions[1].value, tool_parameters: {} }
+    const initialValue: any = { tool_id: selectedOptions[1].value, tool_parameters: { operation: undefined } }
 
     if (value[0] === 'mcp' || (value[0] === 'builtin' && selectedOptions[1]?.children && selectedOptions[1].children.length > 1)) {
-      inititalValue.tool_parameters.operation = value?.[2]
+      initialValue.tool_parameters.operation = value?.[2]
     } else if (value[0] === 'custom') {
-      inititalValue.tool_parameters.operation = selectedOptions?.[2].method_id
+      initialValue.tool_parameters.operation = selectedOptions?.[2].method_id
     }
     curParameters.forEach(vo => {
-      inititalValue.tool_parameters[vo.name] = vo.default
+      initialValue.tool_parameters[vo.name] = vo.default
     })
 
-    form.setFieldsValue(inititalValue)
+    form.setFieldsValue(initialValue)
   }
 
   // string -> string
@@ -244,8 +244,8 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({
               {parameter.type === 'string' && parameter.enum && parameter.enum.length > 0
                 ? <Select size="small" options={parameter.enum.map(vo => ({ value: vo, label: vo }))} placeholder={t('common.pleaseSelect')} />
                 : parameter.type === 'boolean'
-                ? <Switch size="small" />
+                  ? <Switch size="small" />
-                : <Editor
+                  : <Editor
                     variant="outlined"
                     type="input"
                     size="small"