Release/v1.7.8.1

.editorconfig

@@ -0,0 +1,14 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+insert_final_newline = true
+trim_trailing_whitespace = true
+
+[*.{js,jsx,ts,tsx,json}]
+indent_style = space
+indent_size = 2
+
+[*.md]
+trim_trailing_whitespace = false

.github/workflows/docker-deploy.yml

@@ -15,6 +15,11 @@ on:
         description: 'runner array in json format (e.g. ["ubuntu-latest"] or ["self-hosted"])'
         required: true
         default: '[]'
+      app_version:
+        description: 'Docker image tag to build and deploy (e.g. v1.7.1)'
+        required: true
+        default: 'latest'
+        type: string
 
 jobs:
   build-main:
@@ -23,7 +28,7 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
       - name: Build main application image
-        run: docker build --build-arg MIRROR=https://pypi.tuna.tsinghua.edu.cn/simple --build-arg APT_MIRROR=tsinghua -t nexent/nexent -f make/main/Dockerfile .
+        run: docker build --build-arg MIRROR=https://pypi.tuna.tsinghua.edu.cn/simple --build-arg APT_MIRROR=tsinghua -t nexent/nexent:${{ github.event.inputs.app_version }} -t nexent/nexent -f make/main/Dockerfile .
 
   build-data-process:
     runs-on: ${{ fromJson(inputs.runner_label_json) }}
@@ -47,23 +52,23 @@ jobs:
           GIT_TRACE=1 GIT_CURL_VERBOSE=1 GIT_LFS_LOG=debug git lfs pull
           rm -rf .git .gitattributes
       - name: Build data process image
-        run: docker build --build-arg MIRROR=https://pypi.tuna.tsinghua.edu.cn/simple --build-arg APT_MIRROR=tsinghua -t nexent/nexent-data-process -f make/data_process/Dockerfile .
+        run: docker build --build-arg MIRROR=https://pypi.tuna.tsinghua.edu.cn/simple --build-arg APT_MIRROR=tsinghua -t nexent/nexent-data-process:${{ github.event.inputs.app_version }} -t nexent/nexent-data-process -f make/data_process/Dockerfile .
 
   build-web:
     runs-on: ${{ fromJson(inputs.runner_label_json) }}
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
       - name: Build web frontend image
-        run: docker build --build-arg MIRROR=https://registry.npmmirror.com --build-arg APK_MIRROR=tsinghua -t nexent/nexent-web -f make/web/Dockerfile .
+        run: docker build --build-arg MIRROR=https://registry.npmmirror.com --build-arg APK_MIRROR=tsinghua -t nexent/nexent-web:${{ github.event.inputs.app_version }} -t nexent/nexent-web -f make/web/Dockerfile .
 
   build-docs:
     runs-on: ${{ fromJson(inputs.runner_label_json) }}
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
       - name: Build docs image
-        run: docker build --progress=plain -t nexent/nexent-docs -f make/docs/Dockerfile .
+        run: docker build --progress=plain -t nexent/nexent-docs:${{ github.event.inputs.app_version }} -t nexent/nexent-docs -f make/docs/Dockerfile .
 
   deploy:
     runs-on: ${{ fromJson(inputs.runner_label_json) }}
@@ -76,6 +81,9 @@ jobs:
           rm -rf $HOME/nexent
           mkdir -p $HOME/nexent
           cp -r $GITHUB_WORKSPACE/* $HOME/nexent/
+      - name: Force APP_VERSION to latest in deploy.sh (CI only)
+        run: |
+          sed -i 's/APP_VERSION="$(get_app_version)"/APP_VERSION="${{ github.event.inputs.app_version }}"/' $HOME/nexent/docker/deploy.sh
       - name: Start docs container
         run: |
           docker stop nexent-docs 2>/dev/null || true

backend/config_service.py

@@ -1,7 +1,6 @@
 import uvicorn
 import logging
 import warnings
-import asyncio
 
 from consts.const import APP_VERSION
 
@@ -12,30 +11,14 @@
 
 from apps.config_app import app
 from utils.logging_utils import configure_logging, configure_elasticsearch_logging
-from services.tool_configuration_service import initialize_tools_on_startup
+
 
 configure_logging(logging.INFO)
 configure_elasticsearch_logging()
 logger = logging.getLogger("config_service")
 
 
-async def startup_initialization():
-    """
-    Perform initialization tasks during server startup
-    """
+if __name__ == "__main__":
     logger.info("Starting server initialization...")
     logger.info(f"APP version is: {APP_VERSION}")
-    try:
-        # Initialize tools on startup - service layer handles detailed logging
-        await initialize_tools_on_startup()
-        logger.info("Server initialization completed successfully!")
-
-    except Exception as e:
-        logger.error(f"Server initialization failed: {str(e)}")
-        # Don't raise the exception to allow server to start even if initialization fails
-        logger.warning("Server will continue to start despite initialization issues")
-
-
-if __name__ == "__main__":
-    asyncio.run(startup_initialization())
     uvicorn.run(app, host="0.0.0.0", port=5010, log_level="info")

backend/consts/const.py

@@ -285,4 +285,4 @@ class VectorDatabaseType(str, Enum):
 
 
 # APP Version
-APP_VERSION = "v1.7.8"
+APP_VERSION = "v1.7.8.1"

backend/runtime_service.py

@@ -1,7 +1,6 @@
 import uvicorn
 import logging
 import warnings
-import asyncio
 
 from consts.const import APP_VERSION
 
@@ -12,31 +11,16 @@
 
 from apps.runtime_app import app
 from utils.logging_utils import configure_logging, configure_elasticsearch_logging
-from services.tool_configuration_service import initialize_tools_on_startup
+
 
 configure_logging(logging.INFO)
 configure_elasticsearch_logging()
 logger = logging.getLogger("runtime_service")
 
 
-async def startup_initialization():
-    """
-    Perform initialization tasks during server startup
-    """
+if __name__ == "__main__":
     logger.info("Starting server initialization...")
     logger.info(f"APP version is: {APP_VERSION}")
-    try:
-        # Initialize tools on startup - service layer handles detailed logging
-        await initialize_tools_on_startup()
-        logger.info("Server initialization completed successfully!")
-    except Exception as e:
-        logger.error(f"Server initialization failed: {str(e)}")
-        # Don't raise the exception to allow server to start even if initialization fails
-        logger.warning("Server will continue to start despite initialization issues")
-
-
-if __name__ == "__main__":
-    asyncio.run(startup_initialization())
     uvicorn.run(app, host="0.0.0.0", port=5014, log_level="info")
 
 

backend/services/tool_configuration_service.py

@@ -11,7 +11,7 @@
 import jsonref
 from mcpadapt.smolagents_adapter import _sanitize_function_name
 
-from consts.const import DEFAULT_USER_ID, LOCAL_MCP_SERVER, DATA_PROCESS_SERVICE
+from consts.const import LOCAL_MCP_SERVER, DATA_PROCESS_SERVICE
 from consts.exceptions import MCPConnectionError, ToolExecutionException, NotFoundException
 from consts.model import ToolInstanceInfoRequest, ToolInfo, ToolSourceEnum, ToolValidateRequest
 from database.remote_mcp_db import get_mcp_records_by_tenant, get_mcp_server_by_name_and_tenant
@@ -22,7 +22,6 @@
     update_tool_table_from_scan_tool_list,
     search_last_tool_instance_by_tool_id,
 )
-from database.user_tenant_db import get_all_tenant_ids
 from services.file_management_service import get_llm_model
 from services.vectordatabase_service import get_embedding_model, get_vector_db_core
 from services.tenant_config_service import get_selected_knowledge_list, build_knowledge_name_mapping
@@ -367,71 +366,6 @@ async def list_all_tools(tenant_id: str):
     return formatted_tools
 
 
-async def initialize_tools_on_startup():
-    """
-    Initialize and scan all tools during server startup for all tenants
-
-    This function scans all available tools (local, LangChain, and MCP) 
-    and updates the database with the latest tool information for all tenants.
-    """
-
-    logger.info("Starting tool initialization on server startup...")
-
-    try:
-        # Get all tenant IDs from the database
-        tenant_ids = get_all_tenant_ids()
-
-        if not tenant_ids:
-            logger.warning("No tenants found in database, skipping tool initialization")
-            return
-
-        logger.info(f"Found {len(tenant_ids)} tenants: {tenant_ids}")
-
-        total_tools = 0
-        successful_tenants = 0
-        failed_tenants = []
-
-        # Process each tenant
-        for tenant_id in tenant_ids:
-            try:
-                logger.info(f"Initializing tools for tenant: {tenant_id}")
-
-                # Add timeout to prevent hanging during startup
-                try:
-                    await asyncio.wait_for(
-                        update_tool_list(tenant_id=tenant_id, user_id=DEFAULT_USER_ID),
-                        timeout=60.0  # 60 seconds timeout per tenant
-                    )
-
-                    # Get the count of tools for this tenant
-                    tools_info = query_all_tools(tenant_id)
-                    tenant_tool_count = len(tools_info)
-                    total_tools += tenant_tool_count
-                    successful_tenants += 1
-
-                    logger.info(f"Tenant {tenant_id}: {tenant_tool_count} tools initialized")
-
-                except asyncio.TimeoutError:
-                    logger.error(f"Tool initialization timed out for tenant {tenant_id}")
-                    failed_tenants.append(f"{tenant_id} (timeout)")
-
-            except Exception as e:
-                logger.error(f"Tool initialization failed for tenant {tenant_id}: {str(e)}")
-                failed_tenants.append(f"{tenant_id} (error: {str(e)})")
-
-        # Log final results
-        logger.info("Tool initialization completed!")
-        logger.info(f"Total tools available across all tenants: {total_tools}")
-        logger.info(f"Successfully processed: {successful_tenants}/{len(tenant_ids)} tenants")
-
-        if failed_tenants:
-            logger.warning(f"Failed tenants: {', '.join(failed_tenants)}")
-
-    except Exception as e:
-        logger.error(f"❌ Tool initialization failed: {str(e)}")
-        raise
-
-
 def load_last_tool_config_impl(tool_id: int, tenant_id: str, user_id: str):
     """
     Load the last tool configuration for a given tool ID

backend/utils/llm_utils.py

@@ -18,16 +18,37 @@ def _process_thinking_tokens(
 ) -> bool:
     """
     Process tokens to filter out thinking content between <think> and </think> tags.
+    Handles cases where providers only send a closing tag or mix reasoning_content.
     """
-    if is_thinking:
-        return THINK_END_PATTERN not in new_token
+    # Check for end tag first, as it might appear in the same token as start tag
+    if THINK_END_PATTERN in new_token:
+        # If we were never in think mode, treat everything accumulated so far as reasoning and clear it
+        if not is_thinking:
+            token_join.clear()
+            if callback:
+                callback("")  # clear any previously streamed reasoning content
+
+        # Exit thinking mode and only keep content after </think>
+        _, _, after_end = new_token.partition(THINK_END_PATTERN)
+        is_thinking = False
+        new_token = after_end
+        # Continue processing the remaining content in this token
 
+    # Check for start tag (after processing end tag, in case both are in the same token)
     if THINK_START_PATTERN in new_token:
+        # Drop any content before <think> and switch to thinking mode
+        _, _, after_start = new_token.partition(THINK_START_PATTERN)
+        new_token = after_start
+        is_thinking = True
+
+    if is_thinking:
+        # Still inside thinking content; ignore until we exit
         return True
 
-    token_join.append(new_token)
-    if callback:
-        callback("".join(token_join))
+    if new_token:
+        token_join.append(new_token)
+        if callback:
+            callback("".join(token_join))
 
     return False
 
@@ -46,8 +67,8 @@ def call_llm_for_system_prompt(
 
     llm = OpenAIModel(
         model_id=get_model_name_from_config(llm_model_config) if llm_model_config else "",
-        api_base=llm_model_config.get("base_url", ""),
-        api_key=llm_model_config.get("api_key", ""),
+        api_base=llm_model_config.get("base_url", "") if llm_model_config else "",
+        api_key=llm_model_config.get("api_key", "") if llm_model_config else "",
         temperature=0.3,
         top_p=0.95,
     )
@@ -65,16 +86,38 @@ def call_llm_for_system_prompt(
         current_request = llm.client.chat.completions.create(stream=True, **completion_kwargs)
         token_join: List[str] = []
         is_thinking = False
+        reasoning_content_seen = False
+        content_tokens_seen = 0
         for chunk in current_request:
-            new_token = chunk.choices[0].delta.content
+            delta = chunk.choices[0].delta
+            reasoning_content = getattr(delta, "reasoning_content", None)
+            new_token = delta.content
+
+            # Note: reasoning_content is separate metadata and doesn't affect content filtering
+            # We only filter content based on <think> tags in delta.content
+            if reasoning_content:
+                reasoning_content_seen = True
+                logger.debug("Received reasoning_content (metadata only, not filtering content)")
+
+            # Process content token if it exists
             if new_token is not None:
+                content_tokens_seen += 1
                 is_thinking = _process_thinking_tokens(
                     new_token,
                     is_thinking,
                     token_join,
                     callback,
                 )
-        return "".join(token_join)
+
+        result = "".join(token_join)
+        if not result and content_tokens_seen > 0:
+            logger.warning(
+                "Generated prompt is empty but %d content tokens were processed. "
+                "This suggests all content was filtered out.",
+                content_tokens_seen
+            )
+
+        return result
     except Exception as exc:
         logger.error("Failed to generate prompt from LLM: %s", str(exc))
         raise

doc/docs/.vitepress/config.mts

@@ -1,4 +1,4 @@
-// https://vitepress.dev/reference/site-config
+// https://vitepress.dev/reference/site-config
 import { defineConfig } from "vitepress";
 
 export default defineConfig({
@@ -89,6 +89,8 @@ export default defineConfig({
                 text: "Knowledge Base",
                 link: "/en/user-guide/knowledge-base",
               },
+              { text: "MCP Tools", link: "/en/user-guide/mcp-tools" },
+              { text: "Monitoring & Ops", link: "/en/user-guide/monitor" },
               {
                 text: "Model Management",
                 link: "/en/user-guide/model-management",
@@ -267,7 +269,7 @@ export default defineConfig({
             items: [
               { text: "首页", link: "/zh/user-guide/home-page" },
               { text: "开始问答", link: "/zh/user-guide/start-chat" },
-              { text: "快速设置", link: "/zh/user-guide/quick-setup" },
+              { text: "快速配置", link: "/zh/user-guide/quick-setup" },
               { text: "智能体空间", link: "/zh/user-guide/agent-space" },
               { text: "智能体市场", link: "/zh/user-guide/agent-market" },
               {
@@ -278,6 +280,8 @@ export default defineConfig({
                 text: "知识库",
                 link: "/zh/user-guide/knowledge-base",
               },
+              { text: "MCP工具", link: "/zh/user-guide/mcp-tools" },
+              { text: "监控与运维", link: "/zh/user-guide/monitor" },
               { text: "模型管理", link: "/zh/user-guide/model-management" },
               { text: "记忆管理", link: "/zh/user-guide/memory-management" },
               { text: "用户管理", link: "/zh/user-guide/user-management" },