fix: update getServerInfos handling to avoid stale references in smart routing service

feat: add progressive disclosure feature for smart routing tools to reduce token usage
fix: update readonlyAllowPaths for correct authorization handling (#550 )
2026-01-06 14:47:56 -05:00 · 2026-01-05 13:25:27 +08:00 · 2026-01-05 11:15:38 +08:00 · 2026-01-04 13:06:27 +08:00 · 2026-01-03 10:20:17 +08:00 · 2026-01-02 13:42:00 +03:00
42 changed files with 2175 additions and 772 deletions
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -106,7 +106,7 @@ jobs:

  #     - name: Run integration tests
  #       run: |
-  #         export DATABASE_URL="postgresql://postgres:postgres@localhost:5432/mcphub_test"
+  #         export DB_URL="postgresql://postgres:postgres@localhost:5432/mcphub_test"
  #         node test-integration.ts
  #       env:
  #         NODE_ENV: test
--- a/docs/configuration/docker-setup.mdx
+++ b/docs/configuration/docker-setup.mdx
@@ -106,7 +106,7 @@ services:
      - NODE_ENV=production
      - PORT=3000
      - JWT_SECRET=${JWT_SECRET:-your-jwt-secret}
-      - DATABASE_URL=postgresql://mcphub:password@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:password@postgres:5432/mcphub
    volumes:
      - ./mcp_settings.json:/app/mcp_settings.json:ro
      - ./servers.json:/app/servers.json:ro
@@ -180,7 +180,7 @@ services:
      - PORT=3000
      - JWT_SECRET=${JWT_SECRET}
      - JWT_EXPIRES_IN=${JWT_EXPIRES_IN:-24h}
-      - DATABASE_URL=postgresql://mcphub:${POSTGRES_PASSWORD}@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:${POSTGRES_PASSWORD}@postgres:5432/mcphub
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - REDIS_URL=redis://redis:6379
    volumes:
@@ -293,7 +293,7 @@ services:
    environment:
      - NODE_ENV=development
      - PORT=3000
-      - DATABASE_URL=postgresql://mcphub:password@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:password@postgres:5432/mcphub
    volumes:
      - .:/app
      - /app/node_modules
--- a/docs/configuration/mcp-settings.mdx
+++ b/docs/configuration/mcp-settings.mdx
@@ -259,6 +259,92 @@ MCPHub supports environment variable substitution using `${VAR_NAME}` syntax:
 }
 ```

+### Proxy Configuration (proxychains4)
+
+MCPHub supports routing STDIO server network traffic through a proxy using **proxychains4**. This feature is available on **Linux and macOS only** (Windows is not supported).
+
+<Note>
+To use this feature, you must have `proxychains4` installed on your system:
+- **Debian/Ubuntu**: `apt install proxychains4`
+- **macOS**: `brew install proxychains-ng`
+- **Arch Linux**: `pacman -S proxychains-ng`
+</Note>
+
+#### Basic Proxy Configuration
+
+```json
+{
+  "mcpServers": {
+    "fetch-via-proxy": {
+      "command": "uvx",
+      "args": ["mcp-server-fetch"],
+      "proxy": {
+        "enabled": true,
+        "type": "socks5",
+        "host": "127.0.0.1",
+        "port": 1080
+      }
+    }
+  }
+}
+```
+
+#### Proxy Configuration Options
+
+| Field        | Type    | Default   | Description                                      |
+| ------------ | ------- | --------- | ------------------------------------------------ |
+| `enabled`    | boolean | `false`   | Enable/disable proxy routing                     |
+| `type`       | string  | `socks5`  | Proxy protocol: `socks4`, `socks5`, or `http`    |
+| `host`       | string  | -         | Proxy server hostname or IP address              |
+| `port`       | number  | -         | Proxy server port                                |
+| `username`   | string  | -         | Proxy authentication username (optional)         |
+| `password`   | string  | -         | Proxy authentication password (optional)         |
+| `configPath` | string  | -         | Path to custom proxychains4 config file          |
+
+#### Proxy with Authentication
+
+```json
+{
+  "mcpServers": {
+    "secure-server": {
+      "command": "npx",
+      "args": ["-y", "@example/mcp-server"],
+      "proxy": {
+        "enabled": true,
+        "type": "http",
+        "host": "proxy.example.com",
+        "port": 8080,
+        "username": "${PROXY_USER}",
+        "password": "${PROXY_PASSWORD}"
+      }
+    }
+  }
+}
+```
+
+#### Using Custom proxychains4 Configuration
+
+For advanced use cases, you can provide your own proxychains4 configuration file:
+
+```json
+{
+  "mcpServers": {
+    "custom-proxy-server": {
+      "command": "python",
+      "args": ["-m", "custom_mcp_server"],
+      "proxy": {
+        "enabled": true,
+        "configPath": "/etc/proxychains4/custom.conf"
+      }
+    }
+  }
+}
+```
+
+<Tip>
+When `configPath` is specified, all other proxy settings (`type`, `host`, `port`, etc.) are ignored, and the custom configuration file is used directly.
+</Tip>
+
 {/* ### Custom Server Scripts

 #### Local Python Server
--- a/docs/development/getting-started.mdx
+++ b/docs/development/getting-started.mdx
@@ -47,7 +47,7 @@ PORT=3000
 NODE_ENV=development

 # Database Configuration
-DATABASE_URL=postgresql://username:password@localhost:5432/mcphub
+DB_URL=postgresql://username:password@localhost:5432/mcphub

 # JWT Configuration
 JWT_SECRET=your-secret-key
--- a/docs/features/smart-routing.mdx
+++ b/docs/features/smart-routing.mdx
@@ -69,7 +69,7 @@ Smart Routing requires additional setup compared to basic MCPHub usage:
        ports:
          - "3000:3000"
        environment:
-          - DATABASE_URL=postgresql://mcphub:password@postgres:5432/mcphub
+          - DB_URL=postgresql://mcphub:password@postgres:5432/mcphub
          - OPENAI_API_KEY=your_openai_api_key
          - ENABLE_SMART_ROUTING=true
        depends_on:
@@ -114,7 +114,7 @@ Smart Routing requires additional setup compared to basic MCPHub usage:

    2. **Set Environment Variables**:
    ```bash
-    export DATABASE_URL="postgresql://mcphub:your_password@localhost:5432/mcphub"
+    export DB_URL="postgresql://mcphub:your_password@localhost:5432/mcphub"
    export OPENAI_API_KEY="your_openai_api_key"
    export ENABLE_SMART_ROUTING="true"
    ```
@@ -178,7 +178,7 @@ Smart Routing requires additional setup compared to basic MCPHub usage:
          - name: mcphub
            image: samanhappy/mcphub:latest
            env:
-            - name: DATABASE_URL
+            - name: DB_URL
              value: "postgresql://mcphub:password@postgres:5432/mcphub"
            - name: OPENAI_API_KEY
              valueFrom:
@@ -202,7 +202,7 @@ Configure Smart Routing with these environment variables:

 ```bash
 # Required
-DATABASE_URL=postgresql://user:password@host:5432/database
+DB_URL=postgresql://user:password@host:5432/database
 OPENAI_API_KEY=your_openai_api_key

 # Optional
@@ -219,10 +219,10 @@ EMBEDDING_BATCH_SIZE=100
  <Accordion title="Database Configuration">
    ```bash
    # Full PostgreSQL connection string
-    DATABASE_URL=postgresql://username:password@host:port/database?schema=public
+    DB_URL=postgresql://username:password@host:port/database?schema=public

    # SSL configuration for cloud databases
-    DATABASE_URL=postgresql://user:pass@host:5432/db?sslmode=require
+    DB_URL=postgresql://user:pass@host:5432/db?sslmode=require

    # Connection pool settings
    DATABASE_POOL_SIZE=20
@@ -362,6 +362,129 @@ Smart Routing now supports group-scoped searches, allowing you to limit tool dis
  </Accordion>
 </AccordionGroup>

+### Progressive Disclosure Mode
+
+Progressive Disclosure is an optimization feature that reduces token usage when working with Smart Routing. When enabled, the tool discovery workflow changes from a 2-step to a 3-step process.
+
+<AccordionGroup>
+  <Accordion title="What is Progressive Disclosure?">
+    By default, Smart Routing returns full tool information including complete parameter schemas in `search_tools` results. This can consume significant tokens when dealing with tools that have complex input schemas.
+    
+    **Progressive Disclosure** changes this behavior:
+    - `search_tools` returns only tool names and descriptions (minimal info)
+    - A new `describe_tool` endpoint provides full parameter schema on demand
+    - `call_tool` executes the tool as before
+    
+    This approach is particularly useful when:
+    - Working with many tools with complex schemas
+    - Token usage optimization is important
+    - AI clients need to browse many tools before selecting one
+  </Accordion>
+
+  <Accordion title="Enabling Progressive Disclosure">
+    Enable Progressive Disclosure through the Settings page or environment variable:
+    
+    **Via Settings UI:**
+    1. Navigate to Settings → Smart Routing
+    2. Enable the "Progressive Disclosure" toggle
+    3. The change takes effect immediately
+    
+    **Via Environment Variable:**
+    ```bash
+    SMART_ROUTING_PROGRESSIVE_DISCLOSURE=true
+    ```
+  </Accordion>
+
+  <Accordion title="Standard Mode (Default)">
+    When Progressive Disclosure is **disabled** (default), Smart Routing provides two tools:
+    
+    **Workflow:** `search_tools` → `call_tool`
+    
+    | Tool | Purpose |
+    |------|---------|
+    | `search_tools` | Find tools by query, returns full tool info including `inputSchema` |
+    | `call_tool` | Execute a tool with the provided arguments |
+    
+    This mode is simpler but uses more tokens due to full schemas in search results.
+  </Accordion>
+
+  <Accordion title="Progressive Disclosure Mode">
+    When Progressive Disclosure is **enabled**, Smart Routing provides three tools:
+    
+    **Workflow:** `search_tools` → `describe_tool` → `call_tool`
+    
+    | Tool | Purpose |
+    |------|---------|
+    | `search_tools` | Find tools by query, returns only name and description |
+    | `describe_tool` | Get full schema for a specific tool (new) |
+    | `call_tool` | Execute a tool with the provided arguments |
+    
+    **Example workflow:**
+    1. AI calls `search_tools` with query "file operations"
+    2. Results show tool names and descriptions (minimal tokens)
+    3. AI calls `describe_tool` for a specific tool to get full `inputSchema`
+    4. AI calls `call_tool` with the correct arguments
+    
+    This mode reduces token usage by only fetching full schemas when needed.
+  </Accordion>
+
+  <Accordion title="Response Format Comparison">
+    **Standard Mode search_tools response:**
+    ```json
+    {
+      "tools": [
+        {
+          "name": "read_file",
+          "description": "Read contents of a file",
+          "inputSchema": {
+            "type": "object",
+            "properties": {
+              "path": { "type": "string", "description": "File path to read" },
+              "encoding": { "type": "string", "default": "utf-8" }
+            },
+            "required": ["path"]
+          }
+        }
+      ]
+    }
+    ```
+    
+    **Progressive Disclosure Mode search_tools response:**
+    ```json
+    {
+      "tools": [
+        {
+          "name": "read_file",
+          "description": "Read contents of a file"
+        }
+      ],
+      "metadata": {
+        "progressiveDisclosure": true,
+        "guideline": "Use describe_tool to get the full parameter schema before calling."
+      }
+    }
+    ```
+    
+    **describe_tool response:**
+    ```json
+    {
+      "tool": {
+        "name": "read_file",
+        "description": "Read contents of a file",
+        "inputSchema": {
+          "type": "object",
+          "properties": {
+            "path": { "type": "string", "description": "File path to read" },
+            "encoding": { "type": "string", "default": "utf-8" }
+          },
+          "required": ["path"]
+        }
+      }
+    }
+    ```
+  </Accordion>
+</AccordionGroup>
+
 {/* ### Basic Usage

 Connect your AI client to the Smart Routing endpoint and make natural language requests:
@@ -673,11 +796,11 @@ curl -X POST http://localhost:3000/api/smart-routing/feedback \

    **Solutions:**
    1. Verify PostgreSQL is running
-    2. Check DATABASE_URL format
+    2. Check DB_URL format
    3. Ensure pgvector extension is installed
    4. Test connection manually:
    ```bash
-    psql $DATABASE_URL -c "SELECT 1;"
+    psql $DB_URL -c "SELECT 1;"
    ```

  </Accordion>
--- a/docs/installation.mdx
+++ b/docs/installation.mdx
@@ -445,7 +445,7 @@ Set the following environment variables:

 ```bash
 # Database connection
-DATABASE_URL=postgresql://mcphub:your_password@localhost:5432/mcphub
+DB_URL=postgresql://mcphub:your_password@localhost:5432/mcphub

 # OpenAI API for embeddings
 OPENAI_API_KEY=your_openai_api_key
@@ -563,10 +563,10 @@ curl -X POST http://localhost:3000/mcp \
    **Database connection failed:**
    ```bash
    # Test database connection
-    psql $DATABASE_URL -c "SELECT 1;"
+    psql $DB_URL -c "SELECT 1;"

    # Check if pgvector is installed
-    psql $DATABASE_URL -c "CREATE EXTENSION IF NOT EXISTS vector;"
+    psql $DB_URL -c "CREATE EXTENSION IF NOT EXISTS vector;"
    ```

    **Embedding service errors:**
--- a/docs/zh/configuration/docker-setup.mdx
+++ b/docs/zh/configuration/docker-setup.mdx
@@ -106,7 +106,7 @@ services:
      - NODE_ENV=production
      - PORT=3000
      - JWT_SECRET=${JWT_SECRET:-your-jwt-secret}
-      - DATABASE_URL=postgresql://mcphub:password@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:password@postgres:5432/mcphub
    volumes:
      - ./mcp_settings.json:/app/mcp_settings.json:ro
      - ./servers.json:/app/servers.json:ro
@@ -180,7 +180,7 @@ services:
      - PORT=3000
      - JWT_SECRET=${JWT_SECRET}
      - JWT_EXPIRES_IN=${JWT_EXPIRES_IN:-24h}
-      - DATABASE_URL=postgresql://mcphub:${POSTGRES_PASSWORD}@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:${POSTGRES_PASSWORD}@postgres:5432/mcphub
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - REDIS_URL=redis://redis:6379
    volumes:
@@ -293,7 +293,7 @@ services:
    environment:
      - NODE_ENV=development
      - PORT=3000
-      - DATABASE_URL=postgresql://mcphub:password@postgres:5432/mcphub
+      - DB_URL=postgresql://mcphub:password@postgres:5432/mcphub
    volumes:
      - .:/app
      - /app/node_modules
--- a/docs/zh/configuration/mcp-settings.mdx
+++ b/docs/zh/configuration/mcp-settings.mdx
@@ -290,7 +290,7 @@ MCPHub 支持使用 `${VAR_NAME}` 语法进行环境变量替换：
      "command": "python",
      "args": ["-m", "db_server"],
      "env": {
-        "DB_URL": "${NODE_ENV:development == 'production' ? DATABASE_URL : DEV_DATABASE_URL}"
+        "DB_URL": "${NODE_ENV:development == 'production' ? DB_URL : DEV_DB_URL}"
      }
    }
  }
--- a/docs/zh/development/getting-started.mdx
+++ b/docs/zh/development/getting-started.mdx
@@ -47,7 +47,7 @@ PORT=3000
 NODE_ENV=development

 # 数据库配置
-DATABASE_URL=postgresql://username:password@localhost:5432/mcphub
+DB_URL=postgresql://username:password@localhost:5432/mcphub

 # JWT 配置
 JWT_SECRET=your-secret-key
--- a/docs/zh/essentials/code.mdx
+++ b/docs/zh/essentials/code.mdx
@@ -480,7 +480,7 @@ docker run -d \
  --name mcphub \
  -p 3000:3000 \
  -e NODE_ENV=production \
-  -e DATABASE_URL=postgresql://user:pass@host:5432/mcphub \
+  -e DB_URL=postgresql://user:pass@host:5432/mcphub \
  -e JWT_SECRET=your-secret-key \
  mcphub/server:latest

@@ -504,7 +504,7 @@ docker run -d \
  --name mcphub \
  -p 3000:3000 \
  -e NODE_ENV=production \
-  -e DATABASE_URL=postgresql://user:pass@host:5432/mcphub \
+  -e DB_URL=postgresql://user:pass@host:5432/mcphub \
  -e JWT_SECRET=your-secret-key \
  mcphub/server:latest

--- a/docs/zh/essentials/markdown.mdx
+++ b/docs/zh/essentials/markdown.mdx
@@ -159,7 +159,7 @@ services:
      - '3000:3000'
    environment:
      - NODE_ENV=production
-      - DATABASE_URL=postgresql://user:pass@db:5432/mcphub
+      - DB_URL=postgresql://user:pass@db:5432/mcphub
 ```
 ````

@@ -172,7 +172,7 @@ services:
      - '3000:3000'
    environment:
      - NODE_ENV=production
-      - DATABASE_URL=postgresql://user:pass@db:5432/mcphub
+      - DB_URL=postgresql://user:pass@db:5432/mcphub
 ```

 ### 终端命令
--- a/docs/zh/features/smart-routing.mdx
+++ b/docs/zh/features/smart-routing.mdx
@@ -64,17 +64,181 @@ description: '使用向量语义搜索的 AI 工具发现系统'
 <Tabs>
  <Tab title="HTTP MCP">
    ```
+    # 搜索所有服务器
    http://localhost:3000/mcp/$smart
+    
+    # 在特定分组内搜索
+    http://localhost:3000/mcp/$smart/{group}
    ```
  </Tab>

  <Tab title="SSE (Legacy)">
    ```
+    # 搜索所有服务器
    http://localhost:3000/sse/$smart
+    
+    # 在特定分组内搜索
+    http://localhost:3000/sse/$smart/{group}
    ```
  </Tab>
 </Tabs>

+### 分组范围的智能路由
+
+智能路由支持分组范围的搜索，允许您将工具发现限制在特定分组内的服务器：
+
+<AccordionGroup>
+  <Accordion title="使用分组范围的智能路由">
+    将您的 AI 客户端连接到特定分组的智能路由端点：
+    
+    ```
+    http://localhost:3000/mcp/$smart/production
+    ```
+    
+    此端点只会搜索属于 "production" 分组的服务器中的工具。
+    
+    **优势：**
+    - **聚焦结果**：只返回相关服务器的工具
+    - **更好的性能**：减少搜索空间以加快查询速度
+    - **环境隔离**：将开发、预发布和生产工具分开
+    - **访问控制**：根据用户权限限制工具发现
+  </Accordion>
+
+  <Accordion title="工作原理">
+    当使用 `$smart/{group}` 时：
+    
+    1. 系统识别指定的分组
+    2. 获取属于该分组的所有服务器
+    3. 将工具搜索过滤到仅限那些服务器
+    4. 返回限定在该分组服务器范围内的结果
+    
+    如果分组不存在或没有服务器，搜索将不返回任何结果。
+  </Accordion>
+</AccordionGroup>
+
+### 渐进式披露模式
+
+渐进式披露是一个优化功能，可以减少使用智能路由时的 Token 消耗。启用后，工具发现工作流从 2 步变为 3 步。
+
+<AccordionGroup>
+  <Accordion title="什么是渐进式披露？">
+    默认情况下，智能路由在 `search_tools` 结果中返回完整的工具信息，包括完整的参数模式。当处理具有复杂输入模式的工具时，这会消耗大量 Token。
+    
+    **渐进式披露** 改变了这种行为：
+    - `search_tools` 只返回工具名称和描述（最少信息）
+    - 新的 `describe_tool` 端点按需提供完整的参数模式
+    - `call_tool` 像以前一样执行工具
+    
+    这种方法特别适用于：
+    - 处理具有复杂模式的大量工具
+    - Token 使用优化很重要的场景
+    - AI 客户端需要浏览多个工具后再选择
+  </Accordion>
+
+  <Accordion title="启用渐进式披露">
+    通过设置页面或环境变量启用渐进式披露：
+    
+    **通过设置界面：**
+    1. 导航到 设置 → 智能路由
+    2. 启用"渐进式披露"开关
+    3. 更改立即生效
+    
+    **通过环境变量：**
+    ```bash
+    SMART_ROUTING_PROGRESSIVE_DISCLOSURE=true
+    ```
+  </Accordion>
+
+  <Accordion title="标准模式（默认）">
+    当渐进式披露**禁用**（默认）时，智能路由提供两个工具：
+    
+    **工作流：** `search_tools` → `call_tool`
+    
+    | 工具 | 用途 |
+    |------|------|
+    | `search_tools` | 按查询查找工具，返回包含 `inputSchema` 的完整工具信息 |
+    | `call_tool` | 使用提供的参数执行工具 |
+    
+    这种模式更简单，但由于搜索结果中包含完整模式，会使用更多 Token。
+  </Accordion>
+
+  <Accordion title="渐进式披露模式">
+    当渐进式披露**启用**时，智能路由提供三个工具：
+    
+    **工作流：** `search_tools` → `describe_tool` → `call_tool`
+    
+    | 工具 | 用途 |
+    |------|------|
+    | `search_tools` | 按查询查找工具，只返回名称和描述 |
+    | `describe_tool` | 获取特定工具的完整模式（新增） |
+    | `call_tool` | 使用提供的参数执行工具 |
+    
+    **示例工作流：**
+    1. AI 使用查询 "文件操作" 调用 `search_tools`
+    2. 结果显示工具名称和描述（最少 Token）
+    3. AI 为特定工具调用 `describe_tool` 获取完整的 `inputSchema`
+    4. AI 使用正确的参数调用 `call_tool`
+    
+    这种模式通过仅在需要时获取完整模式来减少 Token 使用。
+  </Accordion>
+
+  <Accordion title="响应格式对比">
+    **标准模式 search_tools 响应：**
+    ```json
+    {
+      "tools": [
+        {
+          "name": "read_file",
+          "description": "读取文件内容",
+          "inputSchema": {
+            "type": "object",
+            "properties": {
+              "path": { "type": "string", "description": "要读取的文件路径" },
+              "encoding": { "type": "string", "default": "utf-8" }
+            },
+            "required": ["path"]
+          }
+        }
+      ]
+    }
+    ```
+    
+    **渐进式披露模式 search_tools 响应：**
+    ```json
+    {
+      "tools": [
+        {
+          "name": "read_file",
+          "description": "读取文件内容"
+        }
+      ],
+      "metadata": {
+        "progressiveDisclosure": true,
+        "guideline": "使用 describe_tool 获取完整的参数模式后再调用。"
+      }
+    }
+    ```
+    
+    **describe_tool 响应：**
+    ```json
+    {
+      "tool": {
+        "name": "read_file",
+        "description": "读取文件内容",
+        "inputSchema": {
+          "type": "object",
+          "properties": {
+            "path": { "type": "string", "description": "要读取的文件路径" },
+            "encoding": { "type": "string", "default": "utf-8" }
+          },
+          "required": ["path"]
+        }
+      }
+    }
+    ```
+  </Accordion>
+</AccordionGroup>
+
 {/* ## 性能优化

 ### 嵌入缓存
@@ -245,11 +409,11 @@ curl -X POST http://localhost:3000/api/smart-routing/feedback \

    **解决方案：**
    1. 验证 PostgreSQL 是否正在运行
-    2. 检查 DATABASE_URL 格式
+    2. 检查 DB_URL 格式
    3. 确保安装了 pgvector 扩展
    4. 手动测试连接：
    ```bash
-    psql $DATABASE_URL -c "SELECT 1;"
+    psql $DB_URL -c "SELECT 1;"
    ```

  </Accordion>
--- a/docs/zh/installation.mdx
+++ b/docs/zh/installation.mdx
@@ -420,7 +420,7 @@ description: '各种平台的详细安装说明'

 ```bash
 # 数据库连接
-DATABASE_URL=postgresql://mcphub:your_password@localhost:5432/mcphub
+DB_URL=postgresql://mcphub:your_password@localhost:5432/mcphub

 # 用于嵌入的 OpenAI API
 OPENAI_API_KEY=your_openai_api_key
@@ -538,10 +538,10 @@ curl -X POST http://localhost:3000/mcp \
    **数据库连接失败：**
    ```bash
    # 测试数据库连接
-    psql $DATABASE_URL -c "SELECT 1;"
+    psql $DB_URL -c "SELECT 1;"

    # 检查是否安装了 pgvector
-    psql $DATABASE_URL -c "CREATE EXTENSION IF NOT EXISTS vector;"
+    psql $DB_URL -c "CREATE EXTENSION IF NOT EXISTS vector;"
    ```

    **嵌入服务错误：**
--- a/examples/mcp_settings_with_env_vars.json
+++ b/examples/mcp_settings_with_env_vars.json
@@ -28,7 +28,48 @@
      "env": {
        "API_KEY": "${MY_API_KEY}",
        "DEBUG": "${DEBUG_MODE}",
-        "DATABASE_URL": "${DATABASE_URL}"
+        "DB_URL": "${DB_URL}"
+      }
+    },
+    "example-stdio-with-proxy": {
+      "type": "stdio",
+      "command": "uvx",
+      "args": [
+        "mcp-server-fetch"
+      ],
+      "proxy": {
+        "enabled": true,
+        "type": "socks5",
+        "host": "${PROXY_HOST}",
+        "port": 1080
+      }
+    },
+    "example-stdio-with-auth-proxy": {
+      "type": "stdio",
+      "command": "npx",
+      "args": [
+        "-y",
+        "@example/mcp-server"
+      ],
+      "proxy": {
+        "enabled": true,
+        "type": "http",
+        "host": "${HTTP_PROXY_HOST}",
+        "port": 8080,
+        "username": "${PROXY_USERNAME}",
+        "password": "${PROXY_PASSWORD}"
+      }
+    },
+    "example-stdio-with-custom-proxy-config": {
+      "type": "stdio",
+      "command": "python",
+      "args": [
+        "-m",
+        "custom_mcp_server"
+      ],
+      "proxy": {
+        "enabled": true,
+        "configPath": "/etc/proxychains4/custom.conf"
      }
    },
    "example-openapi-server": {
@@ -55,7 +96,10 @@
        "clientId": "${OAUTH_CLIENT_ID}",
        "clientSecret": "${OAUTH_CLIENT_SECRET}",
        "accessToken": "${OAUTH_ACCESS_TOKEN}",
-        "scopes": ["read", "write"]
+        "scopes": [
+          "read",
+          "write"
+        ]
      }
    }
  },
@@ -77,4 +121,4 @@
      "baseUrl": "${MCPROUTER_BASE_URL}"
    }
  }
-}
+}
--- a/frontend/src/components/GroupCard.tsx
+++ b/frontend/src/components/GroupCard.tsx
@@ -1,99 +1,103 @@
-import { useState, useRef, useEffect } from 'react'
-import { useTranslation } from 'react-i18next'
-import { Group, Server, IGroupServerConfig } from '@/types'
-import { Edit, Trash, Copy, Check, Link, FileCode, DropdownIcon, Wrench } from '@/components/icons/LucideIcons'
-import DeleteDialog from '@/components/ui/DeleteDialog'
-import { useToast } from '@/contexts/ToastContext'
-import { useSettingsData } from '@/hooks/useSettingsData'
+import { useState, useRef, useEffect } from 'react';
+import { useTranslation } from 'react-i18next';
+import { Group, Server, IGroupServerConfig } from '@/types';
+import {
+  Edit,
+  Trash,
+  Copy,
+  Check,
+  Link,
+  FileCode,
+  DropdownIcon,
+  Wrench,
+} from '@/components/icons/LucideIcons';
+import DeleteDialog from '@/components/ui/DeleteDialog';
+import { useToast } from '@/contexts/ToastContext';
+import { useSettingsData } from '@/hooks/useSettingsData';

 interface GroupCardProps {
-  group: Group
-  servers: Server[]
-  onEdit: (group: Group) => void
-  onDelete: (groupId: string) => void
+  group: Group;
+  servers: Server[];
+  onEdit: (group: Group) => void;
+  onDelete: (groupId: string) => void;
 }

-const GroupCard = ({
-  group,
-  servers,
-  onEdit,
-  onDelete
-}: GroupCardProps) => {
-  const { t } = useTranslation()
-  const { showToast } = useToast()
-  const { installConfig } = useSettingsData()
-  const [showDeleteDialog, setShowDeleteDialog] = useState(false)
-  const [copied, setCopied] = useState(false)
-  const [showCopyDropdown, setShowCopyDropdown] = useState(false)
-  const [expandedServer, setExpandedServer] = useState<string | null>(null)
-  const dropdownRef = useRef<HTMLDivElement>(null)
+const GroupCard = ({ group, servers, onEdit, onDelete }: GroupCardProps) => {
+  const { t } = useTranslation();
+  const { showToast } = useToast();
+  const { installConfig } = useSettingsData();
+  const [showDeleteDialog, setShowDeleteDialog] = useState(false);
+  const [copied, setCopied] = useState(false);
+  const [showCopyDropdown, setShowCopyDropdown] = useState(false);
+  const [expandedServer, setExpandedServer] = useState<string | null>(null);
+  const dropdownRef = useRef<HTMLDivElement>(null);

  // Close dropdown when clicking outside
  useEffect(() => {
    const handleClickOutside = (event: MouseEvent) => {
      if (dropdownRef.current && !dropdownRef.current.contains(event.target as Node)) {
-        setShowCopyDropdown(false)
+        setShowCopyDropdown(false);
      }
-    }
+    };

-    document.addEventListener('mousedown', handleClickOutside)
+    document.addEventListener('mousedown', handleClickOutside);
    return () => {
-      document.removeEventListener('mousedown', handleClickOutside)
-    }
-  }, [])
+      document.removeEventListener('mousedown', handleClickOutside);
+    };
+  }, []);

  const handleEdit = () => {
-    onEdit(group)
-  }
+    onEdit(group);
+  };

  const handleDelete = () => {
-    setShowDeleteDialog(true)
-  }
+    setShowDeleteDialog(true);
+  };

  const handleConfirmDelete = () => {
-    onDelete(group.id)
-    setShowDeleteDialog(false)
-  }
+    onDelete(group.id);
+    setShowDeleteDialog(false);
+  };

  const copyToClipboard = (text: string) => {
    if (navigator.clipboard && window.isSecureContext) {
      navigator.clipboard.writeText(text).then(() => {
-        setCopied(true)
-        setShowCopyDropdown(false)
-        showToast(t('common.copySuccess'), 'success')
-        setTimeout(() => setCopied(false), 2000)
-      })
+        setCopied(true);
+        setShowCopyDropdown(false);
+        showToast(t('common.copySuccess'), 'success');
+        setTimeout(() => setCopied(false), 2000);
+      });
    } else {
      // Fallback for HTTP or unsupported clipboard API
-      const textArea = document.createElement('textarea')
-      textArea.value = text
+      const textArea = document.createElement('textarea');
+      textArea.value = text;
      // Avoid scrolling to bottom
-      textArea.style.position = 'fixed'
-      textArea.style.left = '-9999px'
-      document.body.appendChild(textArea)
-      textArea.focus()
-      textArea.select()
+      textArea.style.position = 'fixed';
+      textArea.style.left = '-9999px';
+      document.body.appendChild(textArea);
+      textArea.focus();
+      textArea.select();
      try {
-        document.execCommand('copy')
-        setCopied(true)
-        setShowCopyDropdown(false)
-        showToast(t('common.copySuccess'), 'success')
-        setTimeout(() => setCopied(false), 2000)
+        document.execCommand('copy');
+        setCopied(true);
+        setShowCopyDropdown(false);
+        showToast(t('common.copySuccess'), 'success');
+        setTimeout(() => setCopied(false), 2000);
      } catch (err) {
-        showToast(t('common.copyFailed') || 'Copy failed', 'error')
-        console.error('Copy to clipboard failed:', err)
+        showToast(t('common.copyFailed') || 'Copy failed', 'error');
+        console.error('Copy to clipboard failed:', err);
      }
-      document.body.removeChild(textArea)
+      document.body.removeChild(textArea);
    }
-  }
+  };

  const handleCopyId = () => {
-    copyToClipboard(group.id)
-  }
+    copyToClipboard(group.id);
+  };

  const handleCopyUrl = () => {
-    copyToClipboard(`${installConfig.baseUrl}/mcp/${group.id}`)
-  }
+    copyToClipboard(`${installConfig.baseUrl}/mcp/${group.id}`);
+  };

  const handleCopyJson = () => {
    const jsonConfig = {
@@ -101,23 +105,23 @@ const GroupCard = ({
        mcphub: {
          url: `${installConfig.baseUrl}/mcp/${group.id}`,
          headers: {
-            Authorization: "Bearer <your-access-token>"
-          }
-        }
-      }
-    }
-    copyToClipboard(JSON.stringify(jsonConfig, null, 2))
-  }
+            Authorization: 'Bearer <your-access-token>',
+          },
+        },
+      },
+    };
+    copyToClipboard(JSON.stringify(jsonConfig, null, 2));
+  };

  // Helper function to normalize group servers to get server names
  const getServerNames = (servers: string[] | IGroupServerConfig[]): string[] => {
-    return servers.map(server => typeof server === 'string' ? server : server.name);
+    return servers.map((server) => (typeof server === 'string' ? server : server.name));
  };

  // Helper function to get server configuration
  const getServerConfig = (serverName: string): IGroupServerConfig | undefined => {
-    const server = group.servers.find(s =>
-      typeof s === 'string' ? s === serverName : s.name === serverName
+    const server = group.servers.find((s) =>
+      typeof s === 'string' ? s === serverName : s.name === serverName,
    );
    if (typeof server === 'string') {
      return { name: server, tools: 'all' };
@@ -127,11 +131,11 @@ const GroupCard = ({

  // Get servers that belong to this group
  const serverNames = getServerNames(group.servers);
-  const groupServers = servers.filter(server => serverNames.includes(server.name));
+  const groupServers = servers.filter((server) => serverNames.includes(server.name));

  return (
-    <div className="bg-white shadow rounded-lg p-6 ">
-      <div className="flex justify-between items-center mb-4">
+    <div className="bg-white shadow rounded-lg p-4">
+      <div className="flex justify-between items-center">
        <div>
          <div className="flex items-center">
            <h2 className="text-xl font-semibold text-gray-800">{group.name}</h2>
@@ -175,9 +179,7 @@ const GroupCard = ({
              </div>
            </div>
          </div>
-          {group.description && (
-            <p className="text-gray-600 text-sm mt-1">{group.description}</p>
-          )}
+          {group.description && <p className="text-gray-600 text-sm mt-1">{group.description}</p>}
        </div>
        <div className="flex items-center space-x-3">
          <div className="bg-blue-50 text-blue-700 px-3 py-1 rounded-full text-sm btn-secondary">
@@ -200,17 +202,19 @@ const GroupCard = ({
        </div>
      </div>

-      <div className="mt-4">
+      <div className="">
        {groupServers.length === 0 ? (
          <p className="text-gray-500 italic">{t('groups.noServers')}</p>
        ) : (
          <div className="flex flex-wrap gap-2">
-            {groupServers.map(server => {
+            {groupServers.map((server) => {
              const serverConfig = getServerConfig(server.name);
-              const hasToolRestrictions = serverConfig && serverConfig.tools !== 'all' && Array.isArray(serverConfig.tools);
-              const toolCount = hasToolRestrictions && Array.isArray(serverConfig?.tools)
-                ? serverConfig.tools.length
-                : (server.tools?.length || 0); // Show total tool count when all tools are selected
+              const hasToolRestrictions =
+                serverConfig && serverConfig.tools !== 'all' && Array.isArray(serverConfig.tools);
+              const toolCount =
+                hasToolRestrictions && Array.isArray(serverConfig?.tools)
+                  ? serverConfig.tools.length
+                  : server.tools?.length || 0; // Show total tool count when all tools are selected

              const isExpanded = expandedServer === server.name;

@@ -219,7 +223,7 @@ const GroupCard = ({
                if (hasToolRestrictions && Array.isArray(serverConfig?.tools)) {
                  return serverConfig.tools;
                } else if (server.tools && server.tools.length > 0) {
-                  return server.tools.map(tool => tool.name);
+                  return server.tools.map((tool) => tool.name);
                }
                return [];
              };
@@ -235,9 +239,15 @@ const GroupCard = ({
                    onClick={handleServerClick}
                  >
                    <span className="font-medium text-gray-700 text-sm">{server.name}</span>
-                    <span className={`inline-block h-2 w-2 rounded-full ${server.status === 'connected' ? 'bg-green-500' :
-                      server.status === 'connecting' ? 'bg-yellow-500' : 'bg-red-500'
-                      }`}></span>
+                    <span
+                      className={`inline-block h-2 w-2 rounded-full ${
+                        server.status === 'connected'
+                          ? 'bg-green-500'
+                          : server.status === 'connecting'
+                            ? 'bg-yellow-500'
+                            : 'bg-red-500'
+                      }`}
+                    ></span>
                    {toolCount > 0 && (
                      <span className="text-xs text-blue-600 bg-blue-100 px-2 py-0.5 rounded flex items-center gap-1">
                        <Wrench size={12} />
@@ -278,7 +288,7 @@ const GroupCard = ({
        isGroup={true}
      />
    </div>
-  )
-}
+  );
+};

-export default GroupCard
+export default GroupCard;
--- a/frontend/src/components/ServerCard.tsx
+++ b/frontend/src/components/ServerCard.tsx
@@ -18,7 +18,14 @@ interface ServerCardProps {
  onReload?: (server: Server) => Promise<boolean>;
 }

-const ServerCard = ({ server, onRemove, onEdit, onToggle, onRefresh, onReload }: ServerCardProps) => {
+const ServerCard = ({
+  server,
+  onRemove,
+  onEdit,
+  onToggle,
+  onRefresh,
+  onReload,
+}: ServerCardProps) => {
  const { t } = useTranslation();
  const { showToast } = useToast();
  const [isExpanded, setIsExpanded] = useState(false);
@@ -232,10 +239,10 @@ const ServerCard = ({ server, onRemove, onEdit, onToggle, onRefresh, onReload }:
  return (
    <>
      <div
-        className={`bg-white shadow rounded-lg p-6 mb-6 page-card transition-all duration-200 ${server.enabled === false ? 'opacity-60' : ''}`}
+        className={`bg-white shadow rounded-lg mb-6 page-card transition-all duration-200 ${server.enabled === false ? 'opacity-60' : ''}`}
      >
        <div
-          className="flex justify-between items-center cursor-pointer"
+          className="flex justify-between items-center cursor-pointer p-4"
          onClick={() => setIsExpanded(!isExpanded)}
        >
          <div className="flex items-center space-x-3">
@@ -385,9 +392,9 @@ const ServerCard = ({ server, onRemove, onEdit, onToggle, onRefresh, onReload }:
        {isExpanded && (
          <>
            {server.tools && (
-              <div className="mt-6">
+              <div className="px-4">
                <h6
-                  className={`font-medium ${server.enabled === false ? 'text-gray-600' : 'text-gray-900'} mb-4`}
+                  className={`font-medium ${server.enabled === false ? 'text-gray-600' : 'text-gray-900'} mb-2`}
                >
                  {t('server.tools')}
                </h6>
@@ -405,9 +412,9 @@ const ServerCard = ({ server, onRemove, onEdit, onToggle, onRefresh, onReload }:
            )}

            {server.prompts && (
-              <div className="mt-6">
+              <div className="px-4 pb-2">
                <h6
-                  className={`font-medium ${server.enabled === false ? 'text-gray-600' : 'text-gray-900'} mb-4`}
+                  className={`font-medium ${server.enabled === false ? 'text-gray-600' : 'text-gray-900'}`}
                >
                  {t('server.prompts')}
                </h6>
--- a/frontend/src/components/ui/Pagination.tsx
+++ b/frontend/src/components/ui/Pagination.tsx
@@ -1,16 +1,20 @@
 import React from 'react';
+import { useTranslation } from 'react-i18next';

 interface PaginationProps {
  currentPage: number;
  totalPages: number;
  onPageChange: (page: number) => void;
+  disabled?: boolean;
 }

 const Pagination: React.FC<PaginationProps> = ({
  currentPage,
  totalPages,
-  onPageChange
+  onPageChange,
+  disabled = false
 }) => {
+  const { t } = useTranslation();
  // Generate page buttons
  const getPageButtons = () => {
    const buttons = [];
@@ -95,26 +99,26 @@ const Pagination: React.FC<PaginationProps> = ({
    <div className="flex justify-center items-center my-6">
      <button
        onClick={() => onPageChange(Math.max(1, currentPage - 1))}
-        disabled={currentPage === 1}
-        className={`px-3 py-1 rounded mr-2 ${currentPage === 1
+        disabled={disabled || currentPage === 1}
+        className={`px-3 py-1 rounded mr-2 ${disabled || currentPage === 1
          ? 'bg-gray-100 text-gray-400 cursor-not-allowed'
          : 'bg-gray-200 hover:bg-gray-300 text-gray-700 btn-secondary'
          }`}
      >
-        &laquo; Prev
+        &laquo; {t('common.previous')}
      </button>

      <div className="flex">{getPageButtons()}</div>

      <button
        onClick={() => onPageChange(Math.min(totalPages, currentPage + 1))}
-        disabled={currentPage === totalPages}
-        className={`px-3 py-1 rounded ml-2 ${currentPage === totalPages
+        disabled={disabled || currentPage === totalPages}
+        className={`px-3 py-1 rounded ml-2 ${disabled || currentPage === totalPages
          ? 'bg-gray-100 text-gray-400 cursor-not-allowed'
          : 'bg-gray-200 hover:bg-gray-300 text-gray-700 btn-secondary'
          }`}
      >
-        Next &raquo;
+        {t('common.next')} &raquo;
      </button>
    </div>
  );
--- a/frontend/src/components/ui/PromptCard.tsx
+++ b/frontend/src/components/ui/PromptCard.tsx
@@ -171,9 +171,9 @@ const PromptCard = ({ prompt, server, onToggle, onDescriptionUpdate }: PromptCar
  };

  return (
-    <div className="bg-white border border-gray-200 shadow rounded-lg p-4 mb-4">
+    <div className="bg-white border border-gray-200 shadow rounded-lg mb-4">
      <div
-        className="flex justify-between items-center cursor-pointer"
+        className="flex justify-between items-center p-2 cursor-pointer"
        onClick={() => setIsExpanded(!isExpanded)}
      >
        <div className="flex-1">
--- a/frontend/src/components/ui/ToolCard.tsx
+++ b/frontend/src/components/ui/ToolCard.tsx
@@ -1,19 +1,27 @@
-import { useState, useCallback, useRef, useEffect } from 'react'
-import { useTranslation } from 'react-i18next'
-import { Tool } from '@/types'
-import { ChevronDown, ChevronRight, Play, Loader, Edit, Check, Copy } from '@/components/icons/LucideIcons'
-import { callTool, ToolCallResult, updateToolDescription } from '@/services/toolService'
-import { useSettingsData } from '@/hooks/useSettingsData'
-import { useToast } from '@/contexts/ToastContext'
-import { Switch } from './ToggleGroup'
-import DynamicForm from './DynamicForm'
-import ToolResult from './ToolResult'
+import { useState, useCallback, useRef, useEffect } from 'react';
+import { useTranslation } from 'react-i18next';
+import { Tool } from '@/types';
+import {
+  ChevronDown,
+  ChevronRight,
+  Play,
+  Loader,
+  Edit,
+  Check,
+  Copy,
+} from '@/components/icons/LucideIcons';
+import { callTool, ToolCallResult, updateToolDescription } from '@/services/toolService';
+import { useSettingsData } from '@/hooks/useSettingsData';
+import { useToast } from '@/contexts/ToastContext';
+import { Switch } from './ToggleGroup';
+import DynamicForm from './DynamicForm';
+import ToolResult from './ToolResult';

 interface ToolCardProps {
-  server: string
-  tool: Tool
-  onToggle?: (toolName: string, enabled: boolean) => void
-  onDescriptionUpdate?: (toolName: string, description: string) => void
+  server: string;
+  tool: Tool;
+  onToggle?: (toolName: string, enabled: boolean) => void;
+  onDescriptionUpdate?: (toolName: string, description: string) => void;
 }

 // Helper to check for "empty" values
@@ -26,165 +34,173 @@ function isEmptyValue(value: any): boolean {
 }

 const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps) => {
-  const { t } = useTranslation()
-  const { showToast } = useToast()
-  const { nameSeparator } = useSettingsData()
-  const [isExpanded, setIsExpanded] = useState(false)
-  const [showRunForm, setShowRunForm] = useState(false)
-  const [isRunning, setIsRunning] = useState(false)
-  const [result, setResult] = useState<ToolCallResult | null>(null)
-  const [isEditingDescription, setIsEditingDescription] = useState(false)
-  const [customDescription, setCustomDescription] = useState(tool.description || '')
-  const descriptionInputRef = useRef<HTMLInputElement>(null)
-  const descriptionTextRef = useRef<HTMLSpanElement>(null)
-  const [textWidth, setTextWidth] = useState<number>(0)
-  const [copiedToolName, setCopiedToolName] = useState(false)
+  const { t } = useTranslation();
+  const { showToast } = useToast();
+  const { nameSeparator } = useSettingsData();
+  const [isExpanded, setIsExpanded] = useState(false);
+  const [showRunForm, setShowRunForm] = useState(false);
+  const [isRunning, setIsRunning] = useState(false);
+  const [result, setResult] = useState<ToolCallResult | null>(null);
+  const [isEditingDescription, setIsEditingDescription] = useState(false);
+  const [customDescription, setCustomDescription] = useState(tool.description || '');
+  const descriptionInputRef = useRef<HTMLInputElement>(null);
+  const descriptionTextRef = useRef<HTMLSpanElement>(null);
+  const [textWidth, setTextWidth] = useState<number>(0);
+  const [copiedToolName, setCopiedToolName] = useState(false);

  // Focus the input when editing mode is activated
  useEffect(() => {
    if (isEditingDescription && descriptionInputRef.current) {
-      descriptionInputRef.current.focus()
+      descriptionInputRef.current.focus();
      // Set input width to match text width
      if (textWidth > 0) {
-        descriptionInputRef.current.style.width = `${textWidth + 20}px` // Add some padding
+        descriptionInputRef.current.style.width = `${textWidth + 20}px`; // Add some padding
      }
    }
-  }, [isEditingDescription, textWidth])
+  }, [isEditingDescription, textWidth]);

  // Measure text width when not editing
  useEffect(() => {
    if (!isEditingDescription && descriptionTextRef.current) {
-      setTextWidth(descriptionTextRef.current.offsetWidth)
+      setTextWidth(descriptionTextRef.current.offsetWidth);
    }
-  }, [isEditingDescription, customDescription])
+  }, [isEditingDescription, customDescription]);

  // Generate a unique key for localStorage based on tool name and server
  const getStorageKey = useCallback(() => {
-    return `mcphub_tool_form_${server ? `${server}_` : ''}${tool.name}`
-  }, [tool.name, server])
+    return `mcphub_tool_form_${server ? `${server}_` : ''}${tool.name}`;
+  }, [tool.name, server]);

  // Clear form data from localStorage
  const clearStoredFormData = useCallback(() => {
-    localStorage.removeItem(getStorageKey())
-  }, [getStorageKey])
+    localStorage.removeItem(getStorageKey());
+  }, [getStorageKey]);

  const handleToggle = (enabled: boolean) => {
    if (onToggle) {
-      onToggle(tool.name, enabled)
+      onToggle(tool.name, enabled);
    }
-  }
+  };

  const handleDescriptionEdit = () => {
-    setIsEditingDescription(true)
-  }
+    setIsEditingDescription(true);
+  };

  const handleDescriptionSave = async () => {
    try {
-      const result = await updateToolDescription(server, tool.name, customDescription)
+      const result = await updateToolDescription(server, tool.name, customDescription);
      if (result.success) {
-        setIsEditingDescription(false)
+        setIsEditingDescription(false);
        if (onDescriptionUpdate) {
-          onDescriptionUpdate(tool.name, customDescription)
+          onDescriptionUpdate(tool.name, customDescription);
        }
      } else {
        // Revert on error
-        setCustomDescription(tool.description || '')
-        console.error('Failed to update tool description:', result.error)
+        setCustomDescription(tool.description || '');
+        console.error('Failed to update tool description:', result.error);
      }
    } catch (error) {
-      console.error('Error updating tool description:', error)
-      setCustomDescription(tool.description || '')
-      setIsEditingDescription(false)
+      console.error('Error updating tool description:', error);
+      setCustomDescription(tool.description || '');
+      setIsEditingDescription(false);
    }
-  }
+  };

  const handleDescriptionChange = (e: React.ChangeEvent<HTMLInputElement>) => {
-    setCustomDescription(e.target.value)
-  }
+    setCustomDescription(e.target.value);
+  };

  const handleDescriptionKeyDown = (e: React.KeyboardEvent<HTMLInputElement>) => {
    if (e.key === 'Enter') {
-      handleDescriptionSave()
+      handleDescriptionSave();
    } else if (e.key === 'Escape') {
-      setCustomDescription(tool.description || '')
-      setIsEditingDescription(false)
+      setCustomDescription(tool.description || '');
+      setIsEditingDescription(false);
    }
-  }
+  };

  const handleCopyToolName = async (e: React.MouseEvent) => {
-    e.stopPropagation()
-    
+    e.stopPropagation();
+
    try {
      if (navigator.clipboard && window.isSecureContext) {
-        await navigator.clipboard.writeText(tool.name)
-        setCopiedToolName(true)
-        showToast(t('common.copySuccess'), 'success')
-        setTimeout(() => setCopiedToolName(false), 2000)
+        await navigator.clipboard.writeText(tool.name);
+        setCopiedToolName(true);
+        showToast(t('common.copySuccess'), 'success');
+        setTimeout(() => setCopiedToolName(false), 2000);
      } else {
        // Fallback for HTTP or unsupported clipboard API
-        const textArea = document.createElement('textarea')
-        textArea.value = tool.name
-        textArea.style.position = 'fixed'
-        textArea.style.left = '-9999px'
-        document.body.appendChild(textArea)
-        textArea.focus()
-        textArea.select()
+        const textArea = document.createElement('textarea');
+        textArea.value = tool.name;
+        textArea.style.position = 'fixed';
+        textArea.style.left = '-9999px';
+        document.body.appendChild(textArea);
+        textArea.focus();
+        textArea.select();
        try {
-          document.execCommand('copy')
-          setCopiedToolName(true)
-          showToast(t('common.copySuccess'), 'success')
-          setTimeout(() => setCopiedToolName(false), 2000)
+          document.execCommand('copy');
+          setCopiedToolName(true);
+          showToast(t('common.copySuccess'), 'success');
+          setTimeout(() => setCopiedToolName(false), 2000);
        } catch (err) {
-          showToast(t('common.copyFailed'), 'error')
-          console.error('Copy to clipboard failed:', err)
+          showToast(t('common.copyFailed'), 'error');
+          console.error('Copy to clipboard failed:', err);
        }
-        document.body.removeChild(textArea)
+        document.body.removeChild(textArea);
      }
    } catch (error) {
-      showToast(t('common.copyFailed'), 'error')
-      console.error('Copy to clipboard failed:', error)
+      showToast(t('common.copyFailed'), 'error');
+      console.error('Copy to clipboard failed:', error);
    }
-  }
+  };

  const handleRunTool = async (arguments_: Record<string, any>) => {
-    setIsRunning(true)
+    setIsRunning(true);
    try {
      // filter empty values
-      arguments_ = Object.fromEntries(Object.entries(arguments_).filter(([_, v]) => !isEmptyValue(v)))
-      const result = await callTool({
-        toolName: tool.name,
-        arguments: arguments_,
-      }, server)
+      arguments_ = Object.fromEntries(
+        Object.entries(arguments_).filter(([_, v]) => !isEmptyValue(v)),
+      );
+      const result = await callTool(
+        {
+          toolName: tool.name,
+          arguments: arguments_,
+        },
+        server,
+      );

-      setResult(result)
+      setResult(result);
      // Clear form data on successful submission
      // clearStoredFormData()
    } catch (error) {
      setResult({
        success: false,
        error: error instanceof Error ? error.message : 'Unknown error occurred',
-      })
+      });
    } finally {
-      setIsRunning(false)
+      setIsRunning(false);
    }
-  }
+  };

  const handleCancelRun = () => {
-    setShowRunForm(false)
+    setShowRunForm(false);
    // Clear form data when cancelled
-    clearStoredFormData()
-    setResult(null)
-  }
+    clearStoredFormData();
+    setResult(null);
+  };

  const handleCloseResult = () => {
-    setResult(null)
-  }
+    setResult(null);
+  };

  return (
-    <div className="bg-white border border-gray-200 shadow rounded-lg p-4 mb-4">
+    <div className="bg-white border border-gray-200 shadow rounded-lg mb-4">
      <div
-        className="flex justify-between items-center cursor-pointer"
-        onClick={() => setIsExpanded(!isExpanded)}
+        className="flex justify-between items-center cursor-pointer p-2"
+        onClick={(e) => {
+          e.stopPropagation();
+          setIsExpanded(!isExpanded);
+        }}
      >
        <div className="flex-1">
          <h3 className="text-lg font-medium text-gray-900 inline-flex items-center">
@@ -194,11 +210,7 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
              onClick={handleCopyToolName}
              title={t('common.copy')}
            >
-              {copiedToolName ? (
-                <Check size={16} className="text-green-500" />
-              ) : (
-                <Copy size={16} />
-              )}
+              {copiedToolName ? <Check size={16} className="text-green-500" /> : <Copy size={16} />}
            </button>
            <span className="ml-2 text-sm font-normal text-gray-600 inline-flex items-center">
              {isEditingDescription ? (
@@ -213,14 +225,14 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
                    onClick={(e) => e.stopPropagation()}
                    style={{
                      minWidth: '100px',
-                      width: textWidth > 0 ? `${textWidth + 20}px` : 'auto'
+                      width: textWidth > 0 ? `${textWidth + 20}px` : 'auto',
                    }}
                  />
                  <button
                    className="ml-2 p-1 text-green-600 hover:text-green-800 cursor-pointer transition-colors"
                    onClick={(e) => {
-                      e.stopPropagation()
-                      handleDescriptionSave()
+                      e.stopPropagation();
+                      handleDescriptionSave();
                    }}
                  >
                    <Check size={16} />
@@ -228,12 +240,14 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
                </>
              ) : (
                <>
-                  <span ref={descriptionTextRef}>{customDescription || t('tool.noDescription')}</span>
+                  <span ref={descriptionTextRef}>
+                    {customDescription || t('tool.noDescription')}
+                  </span>
                  <button
                    className="ml-2 p-1 text-gray-500 hover:text-blue-600 cursor-pointer transition-colors"
                    onClick={(e) => {
-                      e.stopPropagation()
-                      handleDescriptionEdit()
+                      e.stopPropagation();
+                      handleDescriptionEdit();
                    }}
                  >
                    <Edit size={14} />
@@ -244,10 +258,7 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
          </h3>
        </div>
        <div className="flex items-center space-x-2">
-          <div
-            className="flex items-center space-x-2"
-            onClick={(e) => e.stopPropagation()}
-          >
+          <div className="flex items-center space-x-2" onClick={(e) => e.stopPropagation()}>
            <Switch
              checked={tool.enabled ?? true}
              onCheckedChange={handleToggle}
@@ -256,18 +267,14 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
          </div>
          <button
            onClick={(e) => {
-              e.stopPropagation()
-              setIsExpanded(true) // Ensure card is expanded when showing run form
-              setShowRunForm(true)
+              e.stopPropagation();
+              setIsExpanded(true); // Ensure card is expanded when showing run form
+              setShowRunForm(true);
            }}
            className="flex items-center space-x-1 px-3 py-1 text-sm text-blue-600 bg-blue-50 hover:bg-blue-100 rounded-md transition-colors btn-primary"
            disabled={isRunning || !tool.enabled}
          >
-            {isRunning ? (
-              <Loader size={14} className="animate-spin" />
-            ) : (
-              <Play size={14} />
-            )}
+            {isRunning ? <Loader size={14} className="animate-spin" /> : <Play size={14} />}
            <span>{isRunning ? t('tool.running') : t('tool.run')}</span>
          </button>
          <button className="text-gray-400 hover:text-gray-600">
@@ -297,7 +304,9 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
                onCancel={handleCancelRun}
                loading={isRunning}
                storageKey={getStorageKey()}
-                title={t('tool.runToolWithName', { name: tool.name.replace(server + nameSeparator, '') })}
+                title={t('tool.runToolWithName', {
+                  name: tool.name.replace(server + nameSeparator, ''),
+                })}
              />
              {/* Tool Result */}
              {result && (
@@ -307,12 +316,10 @@ const ToolCard = ({ tool, server, onToggle, onDescriptionUpdate }: ToolCardProps
              )}
            </div>
          )}
-
-
        </div>
      )}
    </div>
-  )
-}
+  );
+};

-export default ToolCard
+export default ToolCard;
--- a/frontend/src/contexts/ServerContext.tsx
+++ b/frontend/src/contexts/ServerContext.tsx
@@ -17,6 +17,16 @@ const CONFIG = {
  },
 };

+// Pagination info type
+interface PaginationInfo {
+  page: number;
+  limit: number;
+  total: number;
+  totalPages: number;
+  hasNextPage: boolean;
+  hasPrevPage: boolean;
+}
+
 // Context type definition
 interface ServerContextType {
  servers: Server[];
@@ -24,6 +34,11 @@ interface ServerContextType {
  setError: (error: string | null) => void;
  isLoading: boolean;
  fetchAttempts: number;
+  pagination: PaginationInfo | null;
+  currentPage: number;
+  serversPerPage: number;
+  setCurrentPage: (page: number) => void;
+  setServersPerPage: (limit: number) => void;
  triggerRefresh: () => void;
  refreshIfNeeded: () => void; // Smart refresh with debounce
  handleServerAdd: () => void;
@@ -45,6 +60,9 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
  const [refreshKey, setRefreshKey] = useState(0);
  const [isInitialLoading, setIsInitialLoading] = useState(true);
  const [fetchAttempts, setFetchAttempts] = useState(0);
+  const [pagination, setPagination] = useState<PaginationInfo | null>(null);
+  const [currentPage, setCurrentPage] = useState(1);
+  const [serversPerPage, setServersPerPage] = useState(10);

  // Timer reference for polling
  const intervalRef = useRef<NodeJS.Timeout | null>(null);
@@ -73,18 +91,31 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
      const fetchServers = async () => {
        try {
          console.log('[ServerContext] Fetching servers from API...');
-          const data = await apiGet('/servers');
+          // Build query parameters for pagination
+          const params = new URLSearchParams();
+          params.append('page', currentPage.toString());
+          params.append('limit', serversPerPage.toString());
+          const data = await apiGet(`/servers?${params.toString()}`);

          // Update last fetch time
          lastFetchTimeRef.current = Date.now();

          if (data && data.success && Array.isArray(data.data)) {
            setServers(data.data);
+            // Update pagination info if available
+            if (data.pagination) {
+              setPagination(data.pagination);
+            } else {
+              setPagination(null);
+            }
          } else if (data && Array.isArray(data)) {
+            // Compatibility handling for non-paginated responses
            setServers(data);
+            setPagination(null);
          } else {
            console.error('Invalid server data format:', data);
            setServers([]);
+            setPagination(null);
          }

          // Reset error state
@@ -114,7 +145,7 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
      // Set up regular polling
      intervalRef.current = setInterval(fetchServers, CONFIG.normal.pollingInterval);
    },
-    [t],
+    [t, currentPage, serversPerPage],
  );

  // Watch for authentication status changes
@@ -150,7 +181,11 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
    const fetchInitialData = async () => {
      try {
        console.log('[ServerContext] Initial fetch - attempt', attemptsRef.current + 1);
-        const data = await apiGet('/servers');
+        // Build query parameters for pagination
+        const params = new URLSearchParams();
+        params.append('page', currentPage.toString());
+        params.append('limit', serversPerPage.toString());
+        const data = await apiGet(`/servers?${params.toString()}`);

        // Update last fetch time
        lastFetchTimeRef.current = Date.now();
@@ -158,6 +193,12 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
        // Handle API response wrapper object, extract data field
        if (data && data.success && Array.isArray(data.data)) {
          setServers(data.data);
+          // Update pagination info if available
+          if (data.pagination) {
+            setPagination(data.pagination);
+          } else {
+            setPagination(null);
+          }
          setIsInitialLoading(false);
          // Initialization successful, start normal polling (skip immediate to avoid duplicate fetch)
          startNormalPolling({ immediate: false });
@@ -165,6 +206,7 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
        } else if (data && Array.isArray(data)) {
          // Compatibility handling, if API directly returns array
          setServers(data);
+          setPagination(null);
          setIsInitialLoading(false);
          // Initialization successful, start normal polling (skip immediate to avoid duplicate fetch)
          startNormalPolling({ immediate: false });
@@ -173,6 +215,7 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
          // If data format is not as expected, set to empty array
          console.error('Invalid server data format:', data);
          setServers([]);
+          setPagination(null);
          setIsInitialLoading(false);
          // Initialization successful but data is empty, start normal polling (skip immediate)
          startNormalPolling({ immediate: false });
@@ -227,7 +270,7 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
    return () => {
      clearTimer();
    };
-  }, [refreshKey, t, isInitialLoading, startNormalPolling]);
+  }, [refreshKey, t, isInitialLoading, startNormalPolling, currentPage, serversPerPage]);

  // Manually trigger refresh (always refreshes)
  const triggerRefresh = useCallback(() => {
@@ -383,12 +426,28 @@ export const ServerProvider: React.FC<{ children: React.ReactNode }> = ({ childr
    [t, triggerRefresh],
  );

+  // Handle page change
+  const handlePageChange = useCallback((page: number) => {
+    setCurrentPage(page);
+  }, []);
+
+  // Handle servers per page change
+  const handleServersPerPageChange = useCallback((limit: number) => {
+    setServersPerPage(limit);
+    setCurrentPage(1); // Reset to first page when changing page size
+  }, []);
+
  const value: ServerContextType = {
    servers,
    error,
    setError,
    isLoading: isInitialLoading,
    fetchAttempts,
+    pagination,
+    currentPage,
+    serversPerPage,
+    setCurrentPage: handlePageChange,
+    setServersPerPage: handleServersPerPageChange,
    triggerRefresh,
    refreshIfNeeded,
    handleServerAdd,
--- a/frontend/src/contexts/SettingsContext.tsx
+++ b/frontend/src/contexts/SettingsContext.tsx
@@ -33,6 +33,7 @@ interface SmartRoutingConfig {
  openaiApiBaseUrl: string;
  openaiApiKey: string;
  openaiApiEmbeddingModel: string;
+  progressiveDisclosure: boolean;
 }

 interface MCPRouterConfig {
@@ -180,6 +181,7 @@ export const SettingsProvider: React.FC<SettingsProviderProps> = ({ children })
    openaiApiBaseUrl: '',
    openaiApiKey: '',
    openaiApiEmbeddingModel: '',
+    progressiveDisclosure: false,
  });

  const [mcpRouterConfig, setMCPRouterConfig] = useState<MCPRouterConfig>({
@@ -238,6 +240,7 @@ export const SettingsProvider: React.FC<SettingsProviderProps> = ({ children })
          openaiApiKey: data.data.systemConfig.smartRouting.openaiApiKey || '',
          openaiApiEmbeddingModel:
            data.data.systemConfig.smartRouting.openaiApiEmbeddingModel || '',
+          progressiveDisclosure: data.data.systemConfig.smartRouting.progressiveDisclosure ?? false,
        });
      }
      if (data.success && data.data?.systemConfig?.mcpRouter) {
--- a/frontend/src/pages/ServersPage.tsx
+++ b/frontend/src/pages/ServersPage.tsx
@@ -8,6 +8,7 @@ import EditServerForm from '@/components/EditServerForm';
 import { useServerData } from '@/hooks/useServerData';
 import DxtUploadForm from '@/components/DxtUploadForm';
 import JSONImportForm from '@/components/JSONImportForm';
+import Pagination from '@/components/ui/Pagination';

 const ServersPage: React.FC = () => {
  const { t } = useTranslation();
@@ -17,6 +18,11 @@ const ServersPage: React.FC = () => {
    error,
    setError,
    isLoading,
+    pagination,
+    currentPage,
+    serversPerPage,
+    setCurrentPage,
+    setServersPerPage,
    handleServerAdd,
    handleServerEdit,
    handleServerRemove,
@@ -151,19 +157,66 @@ const ServersPage: React.FC = () => {
          <p className="text-gray-600">{t('app.noServers')}</p>
        </div>
      ) : (
-        <div className="space-y-6">
-          {servers.map((server, index) => (
-            <ServerCard
-              key={index}
-              server={server}
-              onRemove={handleServerRemove}
-              onEdit={handleEditClick}
-              onToggle={handleServerToggle}
-              onRefresh={triggerRefresh}
-              onReload={handleServerReload}
-            />
-          ))}
-        </div>
+        <>
+          <div className="space-y-6">
+            {servers.map((server, index) => (
+              <ServerCard
+                key={index}
+                server={server}
+                onRemove={handleServerRemove}
+                onEdit={handleEditClick}
+                onToggle={handleServerToggle}
+                onRefresh={triggerRefresh}
+                onReload={handleServerReload}
+              />
+            ))}
+          </div>
+
+          <div className="flex items-center mb-4">
+            <div className="flex-[2] text-sm text-gray-500">
+              {pagination ? (
+                t('common.showing', {
+                  start: (pagination.page - 1) * pagination.limit + 1,
+                  end: Math.min(pagination.page * pagination.limit, pagination.total),
+                  total: pagination.total
+                })
+              ) : (
+                t('common.showing', {
+                  start: 1,
+                  end: servers.length,
+                  total: servers.length
+                })
+              )}
+            </div>
+            <div className="flex-[4] flex justify-center">
+              {pagination && pagination.totalPages > 1 && (
+                <Pagination
+                  currentPage={currentPage}
+                  totalPages={pagination.totalPages}
+                  onPageChange={setCurrentPage}
+                  disabled={isLoading}
+                />
+              )}
+            </div>
+            <div className="flex-[2] flex items-center justify-end space-x-2">
+              <label htmlFor="perPage" className="text-sm text-gray-600">
+                {t('common.itemsPerPage')}:
+              </label>
+              <select
+                id="perPage"
+                value={serversPerPage}
+                onChange={(e) => setServersPerPage(Number(e.target.value))}
+                disabled={isLoading}
+                className="border rounded p-1 text-sm btn-secondary outline-none disabled:opacity-50 disabled:cursor-not-allowed"
+              >
+                <option value={5}>5</option>
+                <option value={10}>10</option>
+                <option value={20}>20</option>
+                <option value={50}>50</option>
+              </select>
+            </div>
+          </div>
+        </>
      )}

      {editingServer && (
--- a/frontend/src/pages/SettingsPage.tsx
+++ b/frontend/src/pages/SettingsPage.tsx
@@ -1425,6 +1425,24 @@ const SettingsPage: React.FC = () => {
                </div>
              </div>

+              <div className="flex items-center justify-between p-3 bg-gray-50 rounded-md">
+                <div>
+                  <h3 className="font-medium text-gray-700">
+                    {t('settings.progressiveDisclosure')}
+                  </h3>
+                  <p className="text-sm text-gray-500">
+                    {t('settings.progressiveDisclosureDescription')}
+                  </p>
+                </div>
+                <Switch
+                  disabled={loading || !smartRoutingConfig.enabled}
+                  checked={smartRoutingConfig.progressiveDisclosure}
+                  onCheckedChange={(checked) =>
+                    updateSmartRoutingConfig('progressiveDisclosure', checked)
+                  }
+                />
+              </div>
+
              <div className="flex justify-end pt-2">
                <button
                  onClick={handleSaveSmartRoutingConfig}
--- a/frontend/src/types/index.ts
+++ b/frontend/src/types/index.ts
@@ -105,6 +105,17 @@ export interface Prompt {
  enabled?: boolean;
 }

+// Proxychains4 configuration for STDIO servers (Linux/macOS only)
+export interface ProxychainsConfig {
+  enabled?: boolean; // Enable/disable proxychains4 proxy routing
+  type?: 'socks4' | 'socks5' | 'http'; // Proxy protocol type
+  host?: string; // Proxy server hostname or IP address
+  port?: number; // Proxy server port
+  username?: string; // Proxy authentication username (optional)
+  password?: string; // Proxy authentication password (optional)
+  configPath?: string; // Path to custom proxychains4 configuration file (optional)
+}
+
 // Server config types
 export interface ServerConfig {
  type?: 'stdio' | 'sse' | 'streamable-http' | 'openapi';
@@ -123,6 +134,8 @@ export interface ServerConfig {
    resetTimeoutOnProgress?: boolean; // Reset timeout on progress notifications
    maxTotalTimeout?: number; // Maximum total timeout in milliseconds
  }; // MCP request options configuration
+  // Proxychains4 proxy configuration for STDIO servers (Linux/macOS only, Windows not supported)
+  proxy?: ProxychainsConfig;
  // OAuth authentication for upstream MCP servers
  oauth?: {
    clientId?: string; // OAuth client ID
--- a/locales/en.json
+++ b/locales/en.json
@@ -248,6 +248,10 @@
    "wechat": "WeChat",
    "discord": "Discord",
    "required": "Required",
+    "itemsPerPage": "Items per page",
+    "showing": "Showing {{start}}-{{end}} of {{total}}",
+    "previous": "Previous",
+    "next": "Next",
    "secret": "Secret",
    "default": "Default",
    "value": "Value",
@@ -603,6 +607,8 @@
    "openaiApiKeyPlaceholder": "Enter OpenAI API key",
    "openaiApiEmbeddingModel": "OpenAI Embedding Model",
    "openaiApiEmbeddingModelPlaceholder": "text-embedding-3-small",
+    "progressiveDisclosure": "Progressive Disclosure",
+    "progressiveDisclosureDescription": "When enabled, search_tools returns only tool names and descriptions. Use describe_tool to get full parameter schema, reducing token usage.",
    "smartRoutingConfigUpdated": "Smart routing configuration updated successfully",
    "smartRoutingRequiredFields": "Database URL and OpenAI API Key are required to enable smart routing",
    "smartRoutingValidationError": "Please fill in the required fields before enabling Smart Routing: {{fields}}",
--- a/locales/fr.json
+++ b/locales/fr.json
@@ -248,6 +248,10 @@
    "github": "GitHub",
    "wechat": "WeChat",
    "discord": "Discord",
+    "itemsPerPage": "Éléments par page",
+    "showing": "Affichage de {{start}}-{{end}} sur {{total}}",
+    "previous": "Précédent",
+    "next": "Suivant",
    "required": "Requis",
    "secret": "Secret",
    "default": "Défaut",
--- a/locales/tr.json
+++ b/locales/tr.json
@@ -248,6 +248,10 @@
    "github": "GitHub",
    "wechat": "WeChat",
    "discord": "Discord",
+    "itemsPerPage": "Sayfa başına öğe",
+    "showing": "{{total}} öğeden {{start}}-{{end}} gösteriliyor",
+    "previous": "Önceki",
+    "next": "Sonraki",
    "required": "Gerekli",
    "secret": "Gizli",
    "default": "Varsayılan",
--- a/locales/zh.json
+++ b/locales/zh.json
@@ -248,6 +248,10 @@
    "dismiss": "忽略",
    "github": "GitHub",
    "wechat": "微信",
+    "itemsPerPage": "每页显示",
+    "showing": "显示第 {{start}}-{{end}} 条，共 {{total}} 条",
+    "previous": "上一页",
+    "next": "下一页",
    "discord": "Discord",
    "required": "必填",
    "secret": "敏感",
@@ -606,6 +610,8 @@
    "openaiApiKeyPlaceholder": "请输入 OpenAI API 密钥",
    "openaiApiEmbeddingModel": "OpenAI 嵌入模型",
    "openaiApiEmbeddingModelPlaceholder": "text-embedding-3-small",
+    "progressiveDisclosure": "渐进式披露",
+    "progressiveDisclosureDescription": "开启后，search_tools 只返回工具名称和描述，通过 describe_tool 获取完整参数定义，可减少 Token 消耗",
    "smartRoutingConfigUpdated": "智能路由配置更新成功",
    "smartRoutingRequiredFields": "启用智能路由需要填写数据库连接地址和 OpenAI API 密钥",
    "smartRoutingValidationError": "启用智能路由前请先填写必要字段：{{fields}}",
--- a/package.json
+++ b/package.json
@@ -46,7 +46,7 @@
  "license": "ISC",
  "dependencies": {
    "@apidevtools/swagger-parser": "^12.0.0",
-    "@modelcontextprotocol/sdk": "^1.20.2",
+    "@modelcontextprotocol/sdk": "^1.25.1",
    "@node-oauth/oauth2-server": "^5.2.1",
    "@types/adm-zip": "^0.5.7",
    "@types/bcrypt": "^6.0.0",
@@ -57,7 +57,7 @@
    "bcrypt": "^6.0.0",
    "bcryptjs": "^3.0.2",
    "cors": "^2.8.5",
-    "dotenv": "^16.6.1",
+    "dotenv": "^17.2.3",
    "dotenv-expand": "^12.0.2",
    "express": "^4.21.2",
    "express-validator": "^7.3.1",
@@ -108,7 +108,7 @@
    "jest-environment-node": "^30.0.5",
    "jest-mock-extended": "4.0.0",
    "lucide-react": "^0.552.0",
-    "next": "^15.5.0",
+    "next": "^16.1.1",
    "postcss": "^8.5.6",
    "prettier": "^3.6.2",
    "react": "19.2.1",
--- a/pnpm-lock.yaml
+++ b/pnpm-lock.yaml
--- a/src/controllers/serverController.ts
+++ b/src/controllers/serverController.ts
@@ -7,6 +7,7 @@ import {
  BatchCreateServersResponse,
  BatchServerResult,
  ServerConfig,
+  ServerInfo,
 } from '../types/index.js';
 import {
  getServersInfo,
@@ -24,13 +25,66 @@ import { createSafeJSON } from '../utils/serialization.js';
 import { cloneDefaultOAuthServerConfig } from '../constants/oauthServerDefaults.js';
 import { getServerDao, getGroupDao, getSystemConfigDao } from '../dao/DaoFactory.js';
 import { getBearerKeyDao } from '../dao/DaoFactory.js';
+import { UserContextService } from '../services/userContextService.js';

-export const getAllServers = async (_: Request, res: Response): Promise<void> => {
+export const getAllServers = async (req: Request, res: Response): Promise<void> => {
  try {
-    const serversInfo = await getServersInfo();
+    // Parse pagination parameters from query string
+    const page = req.query.page ? parseInt(req.query.page as string, 10) : 1;
+    const limit = req.query.limit ? parseInt(req.query.limit as string, 10) : undefined;
+
+    // Validate pagination parameters
+    if (page < 1) {
+      res.status(400).json({
+        success: false,
+        message: 'Page number must be greater than 0',
+      });
+      return;
+    }
+
+    if (limit !== undefined && (limit < 1 || limit > 1000)) {
+      res.status(400).json({
+        success: false,
+        message: 'Limit must be between 1 and 1000',
+      });
+      return;
+    }
+
+    // Get current user for filtering
+    const currentUser = UserContextService.getInstance().getCurrentUser();
+    const isAdmin = !currentUser || currentUser.isAdmin;
+
+    // Get servers info with pagination if limit is specified
+    let serversInfo: Omit<ServerInfo, 'client' | 'transport'>[];
+    let pagination = undefined;
+
+    if (limit !== undefined) {
+      // Use DAO layer pagination with proper filtering
+      const serverDao = getServerDao();
+      const paginatedResult = isAdmin
+        ? await serverDao.findAllPaginated(page, limit)
+        : await serverDao.findByOwnerPaginated(currentUser!.username, page, limit);
+
+      // Get runtime info for paginated servers
+      serversInfo = await getServersInfo(page, limit, currentUser);
+
+      pagination = {
+        page: paginatedResult.page,
+        limit: paginatedResult.limit,
+        total: paginatedResult.total,
+        totalPages: paginatedResult.totalPages,
+        hasNextPage: paginatedResult.page < paginatedResult.totalPages,
+        hasPrevPage: paginatedResult.page > 1,
+      };
+    } else {
+      // No pagination, get all servers (will be filtered by mcpService)
+      serversInfo = await getServersInfo();
+    }
+
    const response: ApiResponse = {
      success: true,
      data: createSafeJSON(serversInfo),
+      ...(pagination && { pagination }),
    };
    res.json(response);
  } catch (error) {
@@ -564,10 +618,9 @@ export const updateServer = async (req: Request, res: Response): Promise<void> =
      });
    }
  } catch (error) {
-    console.error('Failed to update server:', error);
    res.status(500).json({
      success: false,
-      message: error instanceof Error ? error.message : 'Internal server error',
+      message: 'Internal server error',
    });
  }
 };
@@ -853,7 +906,8 @@ export const updateSystemConfig = async (req: Request, res: Response): Promise<v
        typeof smartRouting.dbUrl === 'string' ||
        typeof smartRouting.openaiApiBaseUrl === 'string' ||
        typeof smartRouting.openaiApiKey === 'string' ||
-        typeof smartRouting.openaiApiEmbeddingModel === 'string');
+        typeof smartRouting.openaiApiEmbeddingModel === 'string' ||
+        typeof smartRouting.progressiveDisclosure === 'boolean');

    const hasMcpRouterUpdate =
      mcpRouter &&
@@ -1064,6 +1118,9 @@ export const updateSystemConfig = async (req: Request, res: Response): Promise<v
      if (typeof smartRouting.openaiApiEmbeddingModel === 'string') {
        systemConfig.smartRouting.openaiApiEmbeddingModel = smartRouting.openaiApiEmbeddingModel;
      }
+      if (typeof smartRouting.progressiveDisclosure === 'boolean') {
+        systemConfig.smartRouting.progressiveDisclosure = smartRouting.progressiveDisclosure;
+      }

      // Check if we need to sync embeddings
      const isNowEnabled = systemConfig.smartRouting.enabled || false;
--- a/src/dao/ServerDao.ts
+++ b/src/dao/ServerDao.ts
@@ -2,10 +2,31 @@ import { ServerConfig } from '../types/index.js';
 import { BaseDao } from './base/BaseDao.js';
 import { JsonFileBaseDao } from './base/JsonFileBaseDao.js';

+/**
+ * Pagination result interface
+ */
+export interface PaginatedResult<T> {
+  data: T[];
+  total: number;
+  page: number;
+  limit: number;
+  totalPages: number;
+}
+
 /**
 * Server DAO interface with server-specific operations
 */
 export interface ServerDao extends BaseDao<ServerConfigWithName, string> {
+  /**
+   * Find all servers with pagination
+   */
+  findAllPaginated(page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>>;
+
+  /**
+   * Find servers by owner with pagination
+   */
+  findByOwnerPaginated(owner: string, page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>>;
+
  /**
   * Find servers by owner
   */
@@ -176,6 +197,61 @@ export class ServerDaoImpl extends JsonFileBaseDao implements ServerDao {
    return servers.length;
  }

+  async findAllPaginated(page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>> {
+    const allServers = await this.getAll();
+    // Sort: enabled servers first, then by creation time
+    const sortedServers = allServers.sort((a, b) => {
+      const aEnabled = a.enabled !== false;
+      const bEnabled = b.enabled !== false;
+      if (aEnabled !== bEnabled) {
+        return aEnabled ? -1 : 1;
+      }
+      return 0; // Keep original order for same enabled status
+    });
+    
+    const total = sortedServers.length;
+    const totalPages = Math.ceil(total / limit);
+    const startIndex = (page - 1) * limit;
+    const endIndex = startIndex + limit;
+    const data = sortedServers.slice(startIndex, endIndex);
+
+    return {
+      data,
+      total,
+      page,
+      limit,
+      totalPages,
+    };
+  }
+
+  async findByOwnerPaginated(owner: string, page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>> {
+    const allServers = await this.getAll();
+    const filteredServers = allServers.filter((server) => server.owner === owner);
+    // Sort: enabled servers first, then by creation time
+    const sortedServers = filteredServers.sort((a, b) => {
+      const aEnabled = a.enabled !== false;
+      const bEnabled = b.enabled !== false;
+      if (aEnabled !== bEnabled) {
+        return aEnabled ? -1 : 1;
+      }
+      return 0; // Keep original order for same enabled status
+    });
+    
+    const total = sortedServers.length;
+    const totalPages = Math.ceil(total / limit);
+    const startIndex = (page - 1) * limit;
+    const endIndex = startIndex + limit;
+    const data = sortedServers.slice(startIndex, endIndex);
+
+    return {
+      data,
+      total,
+      page,
+      limit,
+      totalPages,
+    };
+  }
+
  async findByOwner(owner: string): Promise<ServerConfigWithName[]> {
    const servers = await this.getAll();
    return servers.filter((server) => server.owner === owner);
--- a/src/dao/ServerDaoDbImpl.ts
+++ b/src/dao/ServerDaoDbImpl.ts
@@ -1,4 +1,4 @@
-import { ServerDao, ServerConfigWithName } from './index.js';
+import { ServerDao, ServerConfigWithName, PaginatedResult } from './index.js';
 import { ServerRepository } from '../db/repositories/ServerRepository.js';

 /**
@@ -16,6 +16,32 @@ export class ServerDaoDbImpl implements ServerDao {
    return servers.map((s) => this.mapToServerConfig(s));
  }

+  async findAllPaginated(page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>> {
+    const { data, total } = await this.repository.findAllPaginated(page, limit);
+    const totalPages = Math.ceil(total / limit);
+
+    return {
+      data: data.map((s) => this.mapToServerConfig(s)),
+      total,
+      page,
+      limit,
+      totalPages,
+    };
+  }
+
+  async findByOwnerPaginated(owner: string, page: number, limit: number): Promise<PaginatedResult<ServerConfigWithName>> {
+    const { data, total } = await this.repository.findByOwnerPaginated(owner, page, limit);
+    const totalPages = Math.ceil(total / limit);
+
+    return {
+      data: data.map((s) => this.mapToServerConfig(s)),
+      total,
+      page,
+      limit,
+      totalPages,
+    };
+  }
+
  async findById(name: string): Promise<ServerConfigWithName | null> {
    const server = await this.repository.findByName(name);
    return server ? this.mapToServerConfig(server) : null;
@@ -38,6 +64,7 @@ export class ServerDaoDbImpl implements ServerDao {
      prompts: entity.prompts,
      options: entity.options,
      oauth: entity.oauth,
+      proxy: entity.proxy,
      openapi: entity.openapi,
    });
    return this.mapToServerConfig(server);
@@ -62,6 +89,7 @@ export class ServerDaoDbImpl implements ServerDao {
      prompts: entity.prompts,
      options: entity.options,
      oauth: entity.oauth,
+      proxy: entity.proxy,
      openapi: entity.openapi,
    });
    return server ? this.mapToServerConfig(server) : null;
@@ -140,6 +168,7 @@ export class ServerDaoDbImpl implements ServerDao {
    prompts?: Record<string, { enabled: boolean; description?: string }>;
    options?: Record<string, any>;
    oauth?: Record<string, any>;
+    proxy?: Record<string, any>;
    openapi?: Record<string, any>;
  }): ServerConfigWithName {
    return {
@@ -158,6 +187,7 @@ export class ServerDaoDbImpl implements ServerDao {
      prompts: server.prompts,
      options: server.options,
      oauth: server.oauth,
+      proxy: server.proxy,
      openapi: server.openapi,
    };
  }
--- a/src/db/entities/Server.ts
+++ b/src/db/entities/Server.ts
@@ -59,6 +59,9 @@ export class Server {
  @Column({ type: 'simple-json', nullable: true })
  oauth?: Record<string, any>;

+  @Column({ type: 'simple-json', nullable: true })
+  proxy?: Record<string, any>;
+
  @Column({ type: 'simple-json', nullable: true })
  openapi?: Record<string, any>;

--- a/src/db/repositories/ServerRepository.ts
+++ b/src/db/repositories/ServerRepository.ts
@@ -69,6 +69,41 @@ export class ServerRepository {
    return await this.repository.count();
  }

+  /**
+   * Find servers with pagination
+   */
+  async findAllPaginated(page: number, limit: number): Promise<{ data: Server[]; total: number }> {
+    const skip = (page - 1) * limit;
+    const [data, total] = await this.repository.findAndCount({
+      order: { 
+        enabled: 'DESC',  // Enabled servers first
+        createdAt: 'ASC'  // Then by creation time
+      },
+      skip,
+      take: limit,
+    });
+
+    return { data, total };
+  }
+
+  /**
+   * Find servers by owner with pagination
+   */
+  async findByOwnerPaginated(owner: string, page: number, limit: number): Promise<{ data: Server[]; total: number }> {
+    const skip = (page - 1) * limit;
+    const [data, total] = await this.repository.findAndCount({
+      where: { owner },
+      order: { 
+        enabled: 'DESC',  // Enabled servers first
+        createdAt: 'ASC'  // Then by creation time
+      },
+      skip,
+      take: limit,
+    });
+
+    return { data, total };
+  }
+
  /**
   * Find servers by owner
   */
--- a/src/middlewares/auth.ts
+++ b/src/middlewares/auth.ts
@@ -39,7 +39,7 @@ const validateBearerAuth = async (req: Request): Promise<boolean> => {
  return true;
 };

-const readonlyAllowPaths = ['/tools/call/'];
+const readonlyAllowPaths = ['/tools/'];

 const checkReadonly = (req: Request): boolean => {
  if (!defaultConfig.readonly) {
--- a/src/services/mcpService.ts
+++ b/src/services/mcpService.ts
@@ -1,4 +1,6 @@
 import os from 'os';
+import path from 'path';
+import fs from 'fs';
 import { Server } from '@modelcontextprotocol/sdk/server/index.js';
 import {
  CallToolRequestSchema,
@@ -15,29 +17,187 @@ import {
  StreamableHTTPClientTransportOptions,
 } from '@modelcontextprotocol/sdk/client/streamableHttp.js';
 import { createFetchWithProxy, getProxyConfigFromEnv } from './proxy.js';
-import { ServerInfo, ServerConfig, Tool } from '../types/index.js';
+import { ServerInfo, ServerConfig, Tool, ProxychainsConfig } from '../types/index.js';
 import { expandEnvVars, replaceEnvVars, getNameSeparator } from '../config/index.js';
 import config from '../config/index.js';
 import { getGroup } from './sseService.js';
 import { getServersInGroup, getServerConfigInGroup } from './groupService.js';
-import { saveToolsAsVectorEmbeddings, searchToolsByVector } from './vectorSearchService.js';
+import { saveToolsAsVectorEmbeddings } from './vectorSearchService.js';
 import { OpenAPIClient } from '../clients/openapi.js';
 import { RequestContextService } from './requestContextService.js';
 import { getDataService } from './services.js';
 import { getServerDao, getSystemConfigDao, ServerConfigWithName } from '../dao/index.js';
 import { initializeAllOAuthClients } from './oauthService.js';
 import { createOAuthProvider } from './mcpOAuthProvider.js';
+import {
+  initSmartRoutingService,
+  getSmartRoutingTools,
+  handleSearchToolsRequest,
+  handleDescribeToolRequest,
+  isSmartRoutingGroup,
+} from './smartRoutingService.js';

 const servers: { [sessionId: string]: Server } = {};

 import { setupClientKeepAlive } from './keepAliveService.js';

+/**
+ * Check if proxychains4 is available on the system (Linux/macOS only).
+ * Returns the path to proxychains4 if found, null otherwise.
+ */
+const findProxychains4 = (): string | null => {
+  // Windows is not supported
+  if (process.platform === 'win32') {
+    return null;
+  }
+
+  // Common proxychains4 binary paths
+  const possiblePaths = [
+    '/usr/bin/proxychains4',
+    '/usr/local/bin/proxychains4',
+    '/opt/homebrew/bin/proxychains4', // macOS Homebrew ARM
+    '/usr/local/Cellar/proxychains-ng/*/bin/proxychains4', // macOS Homebrew Intel
+  ];
+
+  for (const p of possiblePaths) {
+    if (fs.existsSync(p)) {
+      return p;
+    }
+  }
+
+  // Try to find in PATH
+  const pathEnv = process.env.PATH || '';
+  const pathDirs = pathEnv.split(path.delimiter);
+  for (const dir of pathDirs) {
+    const fullPath = path.join(dir, 'proxychains4');
+    if (fs.existsSync(fullPath)) {
+      return fullPath;
+    }
+  }
+
+  return null;
+};
+
+/**
+ * Generate a temporary proxychains4 configuration file.
+ * Returns the path to the generated config file.
+ */
+const generateProxychainsConfig = (
+  serverName: string,
+  proxyConfig: ProxychainsConfig,
+): string | null => {
+  // If a custom config path is provided, use it directly
+  if (proxyConfig.configPath) {
+    if (fs.existsSync(proxyConfig.configPath)) {
+      return proxyConfig.configPath;
+    }
+    console.warn(`[${serverName}] Custom proxychains config not found: ${proxyConfig.configPath}`);
+    return null;
+  }
+
+  // Validate required fields
+  if (!proxyConfig.host || !proxyConfig.port) {
+    console.warn(`[${serverName}] Proxy host and port are required for proxychains4`);
+    return null;
+  }
+
+  const proxyType = proxyConfig.type || 'socks5';
+  const proxyLine =
+    proxyConfig.username && proxyConfig.password
+      ? `${proxyType} ${proxyConfig.host} ${proxyConfig.port} ${proxyConfig.username} ${proxyConfig.password}`
+      : `${proxyType} ${proxyConfig.host} ${proxyConfig.port}`;
+
+  const configContent = `# Proxychains4 configuration for MCP server: ${serverName}
+# Generated by MCPHub
+
+localnet 127.0.0.0/255.0.0.0
+localnet 10.0.0.0/255.0.0.0
+localnet 172.16.0.0/255.240.0.0
+localnet 192.168.0.0/255.255.0.0
+
+strict_chain
+proxy_dns
+remote_dns_subnet 224
+tcp_read_time_out 15000
+tcp_connect_time_out 8000
+
+[ProxyList]
+${proxyLine}
+`;
+
+  // Create temp directory if needed
+  const tempDir = path.join(os.tmpdir(), 'mcphub-proxychains');
+  if (!fs.existsSync(tempDir)) {
+    fs.mkdirSync(tempDir, { recursive: true });
+  }
+
+  // Write config file
+  const configPath = path.join(tempDir, `${serverName.replace(/[^a-zA-Z0-9-_]/g, '_')}.conf`);
+  fs.writeFileSync(configPath, configContent, 'utf-8');
+  console.log(`[${serverName}] Generated proxychains4 config: ${configPath}`);
+
+  return configPath;
+};
+
+/**
+ * Wrap a command with proxychains4 if proxy is configured and available.
+ * Returns modified command and args if proxychains4 is used, original values otherwise.
+ */
+const wrapWithProxychains = (
+  serverName: string,
+  command: string,
+  args: string[],
+  proxyConfig?: ProxychainsConfig,
+): { command: string; args: string[] } => {
+  // Skip if proxy is not enabled or not configured
+  if (!proxyConfig?.enabled) {
+    return { command, args };
+  }
+
+  // Check platform - Windows is not supported
+  if (process.platform === 'win32') {
+    console.warn(
+      `[${serverName}] proxychains4 proxy is not supported on Windows, ignoring proxy configuration`,
+    );
+    return { command, args };
+  }
+
+  // Find proxychains4 binary
+  const proxychains4Path = findProxychains4();
+  if (!proxychains4Path) {
+    console.warn(
+      `[${serverName}] proxychains4 not found on system, install it with: apt install proxychains4 (Debian/Ubuntu) or brew install proxychains-ng (macOS)`,
+    );
+    return { command, args };
+  }
+
+  // Generate or get config file
+  const configPath = generateProxychainsConfig(serverName, proxyConfig);
+  if (!configPath) {
+    console.warn(`[${serverName}] Failed to setup proxychains4 configuration, skipping proxy`);
+    return { command, args };
+  }
+
+  // Wrap command with proxychains4
+  console.log(
+    `[${serverName}] Using proxychains4 proxy: ${proxyConfig.type || 'socks5'}://${proxyConfig.host}:${proxyConfig.port}`,
+  );
+
+  return {
+    command: proxychains4Path,
+    args: ['-f', configPath, command, ...args],
+  };
+};
+
 export const initUpstreamServers = async (): Promise<void> => {
  // Initialize OAuth clients for servers with dynamic registration
  await initializeAllOAuthClients();

  // Register all tools from upstream servers
  await registerAllTools(true);
+
+  // Initialize smart routing service with references to mcpService functions
+  initSmartRoutingService(() => serverInfos, filterToolsByConfig, filterToolsByGroup);
 };

 export const getMcpServer = (sessionId?: string, group?: string): Server => {
@@ -209,11 +369,19 @@ export const createTransportFromConfig = async (name: string, conf: ServerConfig
      env['npm_config_registry'] = systemConfig.install.npmRegistry;
    }

-    // Expand environment variables in command
+    // Apply proxychains4 wrapper if proxy is configured (Linux/macOS only)
+    const { command: finalCommand, args: finalArgs } = wrapWithProxychains(
+      name,
+      conf.command,
+      replaceEnvVars(conf.args) as string[],
+      conf.proxy,
+    );
+
+    // Create STDIO transport with potentially wrapped command
    transport = new StdioClientTransport({
      cwd: os.homedir(),
-      command: conf.command,
-      args: replaceEnvVars(conf.args) as string[],
+      command: finalCommand,
+      args: finalArgs,
      env: env,
      stderr: 'pipe',
    });
@@ -618,10 +786,20 @@ export const registerAllTools = async (isInit: boolean, serverName?: string): Pr
 };

 // Get all server information
-export const getServersInfo = async (): Promise<Omit<ServerInfo, 'client' | 'transport'>[]> => {
-  const allServers: ServerConfigWithName[] = await getServerDao().findAll();
+export const getServersInfo = async (
+  page?: number,
+  limit?: number,
+  user?: any,
+): Promise<Omit<ServerInfo, 'client' | 'transport'>[]> => {
  const dataService = getDataService();

+  // Get paginated or all server configurations from DAO
+  // If pagination is used with a non-admin user, filtering is already done at DAO level
+  const isPaginated = limit !== undefined && page !== undefined;
+  const allServers: ServerConfigWithName[] = isPaginated
+    ? (await getServerDao().findAllPaginated(page, limit)).data
+    : await getServerDao().findAll();
+
  // Ensure that servers recently added via DAO but not yet initialized in serverInfos
  // are still visible in the servers list. This avoids a race condition where
  // a POST /api/servers immediately followed by GET /api/servers would not
@@ -629,10 +807,19 @@ export const getServersInfo = async (): Promise<Omit<ServerInfo, 'client' | 'tra
  const combinedServerInfos: ServerInfo[] = [...serverInfos];
  const existingNames = new Set(combinedServerInfos.map((s) => s.name));

+  // Create a set of server names we're interested in (for pagination)
+  const requestedServerNames = new Set(allServers.map((s) => s.name));
+
+  // Filter serverInfos to only include requested servers if pagination is used
+  const filteredServerInfos = isPaginated
+    ? combinedServerInfos.filter((s) => requestedServerNames.has(s.name))
+    : combinedServerInfos;
+
+  // Add servers from DAO that don't have runtime info yet
  for (const server of allServers) {
    if (!existingNames.has(server.name)) {
      const isEnabled = server.enabled === undefined ? true : server.enabled;
-      combinedServerInfos.push({
+      filteredServerInfos.push({
        name: server.name,
        owner: server.owner,
        // Newly created servers that are enabled should appear as "connecting"
@@ -648,12 +835,17 @@ export const getServersInfo = async (): Promise<Omit<ServerInfo, 'client' | 'tra
    }
  }

-  const filterServerInfos: ServerInfo[] = dataService.filterData
-    ? dataService.filterData(combinedServerInfos)
-    : combinedServerInfos;
+  // Apply user filtering only when NOT using pagination (pagination already filtered at DAO level)
+  // Or when no pagination parameters provided (backward compatibility)
+  const shouldApplyUserFilter = !isPaginated;
+  const filterServerInfos: ServerInfo[] =
+    shouldApplyUserFilter && dataService.filterData
+      ? dataService.filterData(filteredServerInfos, user)
+      : filteredServerInfos;

-  const infos = filterServerInfos.map(
-    ({ name, status, tools, prompts, createTime, error, oauth }) => {
+  const infos = filterServerInfos
+    .filter((info) => requestedServerNames.has(info.name)) // Only include requested servers
+    .map(({ name, status, tools, prompts, createTime, error, oauth }) => {
      const serverConfig = allServers.find((server) => server.name === name);
      const enabled = serverConfig ? serverConfig.enabled !== false : true;

@@ -692,12 +884,8 @@ export const getServersInfo = async (): Promise<Omit<ServerInfo, 'client' | 'tra
            }
          : undefined,
      };
-    },
-  );
-  infos.sort((a, b) => {
-    if (a.enabled === b.enabled) return 0;
-    return a.enabled ? -1 : 1;
-  });
+    });
+  // Sorting is now handled at DAO layer for consistent pagination results
  return infos;
 };

@@ -903,89 +1091,10 @@ export const handleListToolsRequest = async (_: any, extra: any) => {
  const group = getGroup(sessionId);
  console.log(`Handling ListToolsRequest for group: ${group}`);

-  // Special handling for $smart group to return special tools
+  // Special handling for $smart group to return smart routing tools
  // Support both $smart and $smart/{group} patterns
-  if (group === '$smart' || group?.startsWith('$smart/')) {
-    // Extract target group if pattern is $smart/{group}
-    const targetGroup = group?.startsWith('$smart/') ? group.substring(7) : undefined;
-
-    // Get info about available servers, filtered by target group if specified
-    let availableServers = serverInfos.filter(
-      (server) => server.status === 'connected' && server.enabled !== false,
-    );
-
-    // If a target group is specified, filter servers to only those in the group
-    if (targetGroup) {
-      const serversInGroup = await getServersInGroup(targetGroup);
-      if (serversInGroup && serversInGroup.length > 0) {
-        availableServers = availableServers.filter((server) =>
-          serversInGroup.includes(server.name),
-        );
-      }
-    }
-
-    // Create simple server information with only server names
-    const serversList = availableServers
-      .map((server) => {
-        return `${server.name}`;
-      })
-      .join(', ');
-
-    const scopeDescription = targetGroup
-      ? `servers in the "${targetGroup}" group`
-      : 'all available servers';
-
-    return {
-      tools: [
-        {
-          name: 'search_tools',
-          description: `STEP 1 of 2: Use this tool FIRST to discover and search for relevant tools across ${scopeDescription}. This tool and call_tool work together as a two-step process: 1) search_tools to find what you need, 2) call_tool to execute it.
-
-For optimal results, use specific queries matching your exact needs. Call this tool multiple times with different queries for different parts of complex tasks. Example queries: "image generation tools", "code review tools", "data analysis", "translation capabilities", etc. Results are sorted by relevance using vector similarity.
-
-After finding relevant tools, you MUST use the call_tool to actually execute them. The search_tools only finds tools - it doesn't execute them.
-
-Available servers: ${serversList}`,
-          inputSchema: {
-            type: 'object',
-            properties: {
-              query: {
-                type: 'string',
-                description:
-                  'The search query to find relevant tools. Be specific and descriptive about the task you want to accomplish.',
-              },
-              limit: {
-                type: 'integer',
-                description:
-                  'Maximum number of results to return. Use higher values (20-30) for broad searches and lower values (5-10) for specific searches.',
-                default: 10,
-              },
-            },
-            required: ['query'],
-          },
-        },
-        {
-          name: 'call_tool',
-          description:
-            "STEP 2 of 2: Use this tool AFTER search_tools to actually execute/invoke any tool you found. This is the execution step - search_tools finds tools, call_tool runs them.\n\nWorkflow: search_tools → examine results → call_tool with the chosen tool name and required arguments.\n\nIMPORTANT: Always check the tool's inputSchema from search_tools results before invoking to ensure you provide the correct arguments. The search results will show you exactly what parameters each tool expects.",
-          inputSchema: {
-            type: 'object',
-            properties: {
-              toolName: {
-                type: 'string',
-                description: 'The exact name of the tool to invoke (from search_tools results)',
-              },
-              arguments: {
-                type: 'object',
-                description:
-                  'The arguments to pass to the tool based on its inputSchema (optional if tool requires no arguments)',
-              },
-            },
-            required: ['toolName'],
-          },
-        },
-      ],
-    };
+  if (isSmartRoutingGroup(group)) {
+    return getSmartRoutingTools(group);
  }

  // Need to filter servers based on group asynchronously
@@ -1037,146 +1146,18 @@ Available servers: ${serversList}`,
 export const handleCallToolRequest = async (request: any, extra: any) => {
  console.log(`Handling CallToolRequest for tool: ${JSON.stringify(request.params)}`);
  try {
-    // Special handling for agent group tools
+    // Special handling for smart routing tools
    if (request.params.name === 'search_tools') {
      const { query, limit = 10 } = request.params.arguments || {};
-
-      if (!query || typeof query !== 'string') {
-        throw new Error('Query parameter is required and must be a string');
-      }
-
-      const limitNum = Math.min(Math.max(parseInt(String(limit)) || 10, 1), 100);
-
-      // Dynamically adjust threshold based on query characteristics
-      let thresholdNum = 0.3; // Default threshold
-
-      // For more general queries, use a lower threshold to get more diverse results
-      if (query.length < 10 || query.split(' ').length <= 2) {
-        thresholdNum = 0.2;
-      }
-
-      // For very specific queries, use a higher threshold for more precise results
-      if (query.length > 30 || query.includes('specific') || query.includes('exact')) {
-        thresholdNum = 0.4;
-      }
-
-      console.log(`Using similarity threshold: ${thresholdNum} for query: "${query}"`);
-
-      // Determine server filtering based on group
      const sessionId = extra.sessionId || '';
-      let group = getGroup(sessionId);
-      let servers: string[] | undefined = undefined; // No server filtering by default
+      return await handleSearchToolsRequest(query, limit, sessionId);
+    }

-      // If group is in format $smart/{group}, filter servers to that group
-      if (group?.startsWith('$smart/')) {
-        const targetGroup = group.substring(7);
-        if (targetGroup) {
-          group = targetGroup;
-        }
-        const serversInGroup = await getServersInGroup(targetGroup);
-        if (serversInGroup !== undefined && serversInGroup !== null) {
-          servers = serversInGroup;
-          if (servers && servers.length > 0) {
-            console.log(
-              `Filtering search to servers in group "${targetGroup}": ${servers.join(', ')}`,
-            );
-          } else {
-            console.log(`Group "${targetGroup}" has no servers, search will return no results`);
-          }
-        }
-      }
-
-      const searchResults = await searchToolsByVector(query, limitNum, thresholdNum, servers);
-      console.log(`Search results: ${JSON.stringify(searchResults)}`);
-      // Find actual tool information from serverInfos by serverName and toolName
-      // First resolve all tool promises
-      const resolvedTools = await Promise.all(
-        searchResults.map(async (result) => {
-          // Find the server in serverInfos
-          const server = serverInfos.find(
-            (serverInfo) =>
-              serverInfo.name === result.serverName &&
-              serverInfo.status === 'connected' &&
-              serverInfo.enabled !== false,
-          );
-          if (server && server.tools && server.tools.length > 0) {
-            // Find the tool in server.tools
-            const actualTool = server.tools.find((tool) => tool.name === result.toolName);
-            if (actualTool) {
-              // Check if the tool is enabled in configuration
-              const tools = await filterToolsByConfig(server.name, [actualTool]);
-              if (tools.length > 0) {
-                // Apply custom description from configuration
-                const serverConfig = await getServerDao().findById(server.name);
-                const toolConfig = serverConfig?.tools?.[actualTool.name];
-
-                // Return the actual tool info from serverInfos with custom description
-                return {
-                  ...actualTool,
-                  description: toolConfig?.description || actualTool.description,
-                  serverName: result.serverName, // Add serverName for filtering
-                };
-              }
-            }
-          }
-
-          // Fallback to search result if server or tool not found or disabled
-          return {
-            name: result.toolName,
-            description: result.description || '',
-            inputSchema: cleanInputSchema(result.inputSchema || {}),
-            serverName: result.serverName, // Add serverName for filtering
-          };
-        }),
-      );
-
-      // Now filter the resolved tools
-      const filterResults = await Promise.all(
-        resolvedTools.map(async (tool) => {
-          if (tool.name) {
-            const serverName = tool.serverName;
-            if (serverName) {
-              let tools = await filterToolsByConfig(serverName, [tool as Tool]);
-              if (tools.length === 0) {
-                return false;
-              }
-
-              tools = await filterToolsByGroup(group, serverName, tools);
-              return tools.length > 0;
-            }
-          }
-          return true;
-        }),
-      );
-      const tools = resolvedTools.filter((_, i) => filterResults[i]);
-
-      // Add usage guidance to the response
-      const response = {
-        tools,
-        metadata: {
-          query: query,
-          threshold: thresholdNum,
-          totalResults: tools.length,
-          guideline:
-            tools.length > 0
-              ? "Found relevant tools. If these tools don't match exactly what you need, try another search with more specific keywords."
-              : 'No tools found. Try broadening your search or using different keywords.',
-          nextSteps:
-            tools.length > 0
-              ? 'To use a tool, call call_tool with the toolName and required arguments.'
-              : 'Consider searching for related capabilities or more general terms.',
-        },
-      };
-
-      // Return in the same format as handleListToolsRequest
-      return {
-        content: [
-          {
-            type: 'text',
-            text: JSON.stringify(response),
-          },
-        ],
-      };
+    // Special handling for describe_tool (progressive disclosure mode)
+    if (request.params.name === 'describe_tool') {
+      const { toolName } = request.params.arguments || {};
+      const sessionId = extra.sessionId || '';
+      return await handleDescribeToolRequest(toolName, sessionId);
    }

    // Special handling for call_tool
--- a/src/services/smartRoutingService.ts
+++ b/src/services/smartRoutingService.ts
@@ -0,0 +1,525 @@
+/**
+ * Smart Routing Service
+ *
+ * This service handles the $smart routing functionality, which provides
+ * AI-powered tool discovery using vector semantic search.
+ */
+
+import { Tool, ServerInfo } from '../types/index.js';
+import { getServersInGroup } from './groupService.js';
+import { searchToolsByVector } from './vectorSearchService.js';
+import { getSmartRoutingConfig } from '../utils/smartRouting.js';
+import { getServerDao } from '../dao/index.js';
+import { getGroup } from './sseService.js';
+
+// Reference to serverInfos from mcpService - will be set via init
+let serverInfosRef: ServerInfo[] = [];
+let getServerInfosFn: () => ServerInfo[] = () => serverInfosRef;
+let filterToolsByConfigFn: (serverName: string, tools: Tool[]) => Promise<Tool[]>;
+let filterToolsByGroupFn: (
+  group: string | undefined,
+  serverName: string,
+  tools: Tool[],
+) => Promise<Tool[]>;
+
+/**
+ * Initialize the smart routing service with references to mcpService functions
+ */
+export const initSmartRoutingService = (
+  getServerInfos: () => ServerInfo[],
+  filterToolsByConfig: (serverName: string, tools: Tool[]) => Promise<Tool[]>,
+  filterToolsByGroup: (
+    group: string | undefined,
+    serverName: string,
+    tools: Tool[],
+  ) => Promise<Tool[]>,
+) => {
+  // Store the getter to avoid stale references while staying ESM-safe
+  getServerInfosFn = getServerInfos;
+  serverInfosRef = getServerInfos();
+  filterToolsByConfigFn = filterToolsByConfig;
+  filterToolsByGroupFn = filterToolsByGroup;
+};
+
+/**
+ * Get current server infos (refreshed each call)
+ */
+const getServerInfos = (): ServerInfo[] => {
+  return getServerInfosFn();
+};
+
+/**
+ * Helper function to clean $schema field from inputSchema
+ */
+const cleanInputSchema = (schema: any): any => {
+  if (!schema || typeof schema !== 'object') {
+    return schema;
+  }
+
+  const cleanedSchema = { ...schema };
+  delete cleanedSchema.$schema;
+
+  return cleanedSchema;
+};
+
+/**
+ * Generate the list of smart routing tools based on configuration
+ */
+export const getSmartRoutingTools = async (
+  group: string | undefined,
+): Promise<{ tools: any[] }> => {
+  // Extract target group if pattern is $smart/{group}
+  const targetGroup = group?.startsWith('$smart/') ? group.substring(7) : undefined;
+
+  // Get smart routing config to check progressive disclosure setting
+  const smartRoutingConfig = await getSmartRoutingConfig();
+  const progressiveDisclosure = smartRoutingConfig.progressiveDisclosure ?? false;
+
+  // Get info about available servers, filtered by target group if specified
+  let availableServers = getServerInfos().filter(
+    (server) => server.status === 'connected' && server.enabled !== false,
+  );
+
+  // If a target group is specified, filter servers to only those in the group
+  if (targetGroup) {
+    const serversInGroup = await getServersInGroup(targetGroup);
+    if (serversInGroup && serversInGroup.length > 0) {
+      availableServers = availableServers.filter((server) => serversInGroup.includes(server.name));
+    }
+  }
+
+  // Create simple server information with only server names
+  const serversList = availableServers
+    .map((server) => {
+      return `${server.name}`;
+    })
+    .join(', ');
+
+  const scopeDescription = targetGroup
+    ? `servers in the "${targetGroup}" group`
+    : 'all available servers';
+
+  // Base tools that are always available
+  const tools: any[] = [];
+
+  if (progressiveDisclosure) {
+    // Progressive disclosure mode: search_tools returns minimal info,
+    // describe_tool provides full schema
+    tools.push(
+      {
+        name: 'search_tools',
+        description: `STEP 1 of 3: Use this tool FIRST to discover and search for relevant tools across ${scopeDescription}. Returns tool names and descriptions only - use describe_tool to get full parameter details before calling.
+
+For optimal results, use specific queries matching your exact needs. Call this tool multiple times with different queries for different parts of complex tasks. Example queries: "image generation tools", "code review tools", "data analysis", "translation capabilities", etc. Results are sorted by relevance using vector similarity.
+
+Workflow: search_tools → describe_tool (for parameter details) → call_tool (to execute)
+
+Available servers: ${serversList}`,
+        inputSchema: {
+          type: 'object',
+          properties: {
+            query: {
+              type: 'string',
+              description:
+                'The search query to find relevant tools. Be specific and descriptive about the task you want to accomplish.',
+            },
+            limit: {
+              type: 'integer',
+              description:
+                'Maximum number of results to return. Use higher values (20-30) for broad searches and lower values (5-10) for specific searches.',
+              default: 10,
+            },
+          },
+          required: ['query'],
+        },
+        annotations: {
+          title: 'Search Tools',
+          readOnlyHint: true,
+        },
+      },
+      {
+        name: 'describe_tool',
+        description:
+          'STEP 2 of 3: Use this tool AFTER search_tools to get the full parameter schema for a specific tool. This provides the complete inputSchema needed to correctly invoke the tool with call_tool.\n\nWorkflow: search_tools → describe_tool → call_tool',
+        inputSchema: {
+          type: 'object',
+          properties: {
+            toolName: {
+              type: 'string',
+              description: 'The exact name of the tool to describe (from search_tools results)',
+            },
+          },
+          required: ['toolName'],
+        },
+        annotations: {
+          title: 'Describe Tool',
+          readOnlyHint: true,
+        },
+      },
+      {
+        name: 'call_tool',
+        description:
+          "STEP 3 of 3: Use this tool AFTER describe_tool to actually execute/invoke any tool you found. This is the execution step.\n\nWorkflow: search_tools → describe_tool → call_tool with the chosen tool name and required arguments.\n\nIMPORTANT: Always use describe_tool first to get the tool's inputSchema before invoking to ensure you provide the correct arguments.",
+        inputSchema: {
+          type: 'object',
+          properties: {
+            toolName: {
+              type: 'string',
+              description: 'The exact name of the tool to invoke (from search_tools results)',
+            },
+            arguments: {
+              type: 'object',
+              description:
+                'The arguments to pass to the tool based on its inputSchema from describe_tool (optional if tool requires no arguments)',
+            },
+          },
+          required: ['toolName'],
+        },
+        annotations: {
+          title: 'Call Tool',
+          openWorldHint: true,
+        },
+      },
+    );
+  } else {
+    // Standard mode: search_tools returns full schema
+    tools.push(
+      {
+        name: 'search_tools',
+        description: `STEP 1 of 2: Use this tool FIRST to discover and search for relevant tools across ${scopeDescription}. This tool and call_tool work together as a two-step process: 1) search_tools to find what you need, 2) call_tool to execute it.
+
+For optimal results, use specific queries matching your exact needs. Call this tool multiple times with different queries for different parts of complex tasks. Example queries: "image generation tools", "code review tools", "data analysis", "translation capabilities", etc. Results are sorted by relevance using vector similarity.
+
+After finding relevant tools, you MUST use the call_tool to actually execute them. The search_tools only finds tools - it doesn't execute them.
+
+Available servers: ${serversList}`,
+        inputSchema: {
+          type: 'object',
+          properties: {
+            query: {
+              type: 'string',
+              description:
+                'The search query to find relevant tools. Be specific and descriptive about the task you want to accomplish.',
+            },
+            limit: {
+              type: 'integer',
+              description:
+                'Maximum number of results to return. Use higher values (20-30) for broad searches and lower values (5-10) for specific searches.',
+              default: 10,
+            },
+          },
+          required: ['query'],
+        },
+        annotations: {
+          title: 'Search Tools',
+          readOnlyHint: true,
+        },
+      },
+      {
+        name: 'call_tool',
+        description:
+          "STEP 2 of 2: Use this tool AFTER search_tools to actually execute/invoke any tool you found. This is the execution step - search_tools finds tools, call_tool runs them.\n\nWorkflow: search_tools → examine results → call_tool with the chosen tool name and required arguments.\n\nIMPORTANT: Always check the tool's inputSchema from search_tools results before invoking to ensure you provide the correct arguments. The search results will show you exactly what parameters each tool expects.",
+        inputSchema: {
+          type: 'object',
+          properties: {
+            toolName: {
+              type: 'string',
+              description: 'The exact name of the tool to invoke (from search_tools results)',
+            },
+            arguments: {
+              type: 'object',
+              description:
+                'The arguments to pass to the tool based on its inputSchema (optional if tool requires no arguments)',
+            },
+          },
+          required: ['toolName'],
+        },
+        annotations: {
+          title: 'Call Tool',
+          openWorldHint: true,
+        },
+      },
+    );
+  }
+
+  return { tools };
+};
+
+/**
+ * Handle the search_tools request for smart routing
+ */
+export const handleSearchToolsRequest = async (
+  query: string,
+  limit: number,
+  sessionId: string,
+): Promise<any> => {
+  if (!query || typeof query !== 'string') {
+    throw new Error('Query parameter is required and must be a string');
+  }
+
+  const limitNum = Math.min(Math.max(parseInt(String(limit)) || 10, 1), 100);
+
+  // Dynamically adjust threshold based on query characteristics
+  let thresholdNum = 0.3; // Default threshold
+
+  // For more general queries, use a lower threshold to get more diverse results
+  if (query.length < 10 || query.split(' ').length <= 2) {
+    thresholdNum = 0.2;
+  }
+
+  // For very specific queries, use a higher threshold for more precise results
+  if (query.length > 30 || query.includes('specific') || query.includes('exact')) {
+    thresholdNum = 0.4;
+  }
+
+  console.log(`Using similarity threshold: ${thresholdNum} for query: "${query}"`);
+
+  // Determine server filtering based on group
+  let group = getGroup(sessionId);
+  let servers: string[] | undefined = undefined; // No server filtering by default
+
+  // If group is in format $smart/{group}, filter servers to that group
+  if (group?.startsWith('$smart/')) {
+    const targetGroup = group.substring(7);
+    if (targetGroup) {
+      group = targetGroup;
+    }
+    const serversInGroup = await getServersInGroup(targetGroup);
+    if (serversInGroup !== undefined && serversInGroup !== null) {
+      servers = serversInGroup;
+      if (servers && servers.length > 0) {
+        console.log(`Filtering search to servers in group "${targetGroup}": ${servers.join(', ')}`);
+      } else {
+        console.log(`Group "${targetGroup}" has no servers, search will return no results`);
+      }
+    }
+  }
+
+  const searchResults = await searchToolsByVector(query, limitNum, thresholdNum, servers);
+  console.log(`Search results: ${JSON.stringify(searchResults)}`);
+
+  // Get smart routing config to check progressive disclosure setting
+  const smartRoutingConfig = await getSmartRoutingConfig();
+  const progressiveDisclosure = smartRoutingConfig.progressiveDisclosure ?? false;
+
+  // Find actual tool information from serverInfos by serverName and toolName
+  const resolvedTools = await Promise.all(
+    searchResults.map(async (result) => {
+      // Find the server in serverInfos
+      const server = getServerInfos().find(
+        (serverInfo) =>
+          serverInfo.name === result.serverName &&
+          serverInfo.status === 'connected' &&
+          serverInfo.enabled !== false,
+      );
+      if (server && server.tools && server.tools.length > 0) {
+        // Find the tool in server.tools
+        const actualTool = server.tools.find((tool) => tool.name === result.toolName);
+        if (actualTool) {
+          // Check if the tool is enabled in configuration
+          const tools = await filterToolsByConfigFn(server.name, [actualTool]);
+          if (tools.length > 0) {
+            // Apply custom description from configuration
+            const serverConfig = await getServerDao().findById(server.name);
+            const toolConfig = serverConfig?.tools?.[actualTool.name];
+
+            // Return the actual tool info from serverInfos with custom description
+            if (progressiveDisclosure) {
+              // Progressive disclosure: return only name and description
+              return {
+                name: actualTool.name,
+                description: toolConfig?.description || actualTool.description,
+                serverName: result.serverName,
+              };
+            } else {
+              // Standard mode: return full tool info
+              return {
+                ...actualTool,
+                description: toolConfig?.description || actualTool.description,
+                serverName: result.serverName,
+              };
+            }
+          }
+        }
+      }
+
+      // Fallback to search result if server or tool not found or disabled
+      if (progressiveDisclosure) {
+        return {
+          name: result.toolName,
+          description: result.description || '',
+          serverName: result.serverName,
+        };
+      } else {
+        return {
+          name: result.toolName,
+          description: result.description || '',
+          inputSchema: cleanInputSchema(result.inputSchema || {}),
+          serverName: result.serverName,
+        };
+      }
+    }),
+  );
+
+  // Filter the resolved tools
+  const filterResults = await Promise.all(
+    resolvedTools.map(async (tool) => {
+      if (tool.name) {
+        const serverName = tool.serverName;
+        if (serverName) {
+          let tools = await filterToolsByConfigFn(serverName, [tool as Tool]);
+          if (tools.length === 0) {
+            return false;
+          }
+
+          tools = await filterToolsByGroupFn(group, serverName, tools);
+          return tools.length > 0;
+        }
+      }
+      return true;
+    }),
+  );
+  const tools = resolvedTools.filter((_, i) => filterResults[i]);
+
+  // Build response based on mode
+  let guideline: string;
+  let nextSteps: string;
+
+  if (progressiveDisclosure) {
+    guideline =
+      tools.length > 0
+        ? "Found relevant tools. Use describe_tool to get the full parameter schema before calling. If these tools don't match exactly what you need, try another search with more specific keywords."
+        : 'No tools found. Try broadening your search or using different keywords.';
+    nextSteps =
+      tools.length > 0
+        ? 'Use describe_tool with the toolName to get the full inputSchema, then use call_tool to execute.'
+        : 'Consider searching for related capabilities or more general terms.';
+  } else {
+    guideline =
+      tools.length > 0
+        ? "Found relevant tools. If these tools don't match exactly what you need, try another search with more specific keywords."
+        : 'No tools found. Try broadening your search or using different keywords.';
+    nextSteps =
+      tools.length > 0
+        ? 'To use a tool, call call_tool with the toolName and required arguments.'
+        : 'Consider searching for related capabilities or more general terms.';
+  }
+
+  const response = {
+    tools,
+    metadata: {
+      query: query,
+      threshold: thresholdNum,
+      totalResults: tools.length,
+      progressiveDisclosure,
+      guideline,
+      nextSteps,
+    },
+  };
+
+  return {
+    content: [
+      {
+        type: 'text',
+        text: JSON.stringify(response),
+      },
+    ],
+  };
+};
+
+/**
+ * Handle the describe_tool request for smart routing (progressive disclosure mode)
+ */
+export const handleDescribeToolRequest = async (
+  toolName: string,
+  sessionId: string,
+): Promise<any> => {
+  if (!toolName || typeof toolName !== 'string') {
+    throw new Error('toolName parameter is required and must be a string');
+  }
+
+  console.log(`Handling describe_tool request for: ${toolName}`);
+
+  // Determine group filtering
+  let group = getGroup(sessionId);
+  if (group?.startsWith('$smart/')) {
+    group = group.substring(7);
+  }
+
+  // Find the tool across all connected servers
+  for (const serverInfo of getServerInfos()) {
+    if (serverInfo.status !== 'connected' || serverInfo.enabled === false) {
+      continue;
+    }
+
+    // Check if this server has the tool
+    const tool = serverInfo.tools?.find((t) => t.name === toolName);
+    if (!tool) {
+      continue;
+    }
+
+    // Check if the tool is enabled in configuration
+    const tools = await filterToolsByConfigFn(serverInfo.name, [tool]);
+    if (tools.length === 0) {
+      continue;
+    }
+
+    // Apply group filtering if applicable
+    if (group) {
+      const filteredTools = await filterToolsByGroupFn(group, serverInfo.name, tools);
+      if (filteredTools.length === 0) {
+        continue;
+      }
+    }
+
+    // Get custom description from configuration
+    const serverConfig = await getServerDao().findById(serverInfo.name);
+    const toolConfig = serverConfig?.tools?.[tool.name];
+
+    // Return full tool information
+    const toolInfo = {
+      name: tool.name,
+      description: toolConfig?.description || tool.description,
+      inputSchema: cleanInputSchema(tool.inputSchema),
+      serverName: serverInfo.name,
+    };
+
+    return {
+      content: [
+        {
+          type: 'text',
+          text: JSON.stringify({
+            tool: toolInfo,
+            metadata: {
+              message: `Full schema for tool '${toolName}'. Use call_tool with the toolName and arguments based on the inputSchema.`,
+            },
+          }),
+        },
+      ],
+    };
+  }
+
+  // Tool not found
+  return {
+    content: [
+      {
+        type: 'text',
+        text: JSON.stringify({
+          error: `Tool '${toolName}' not found or not available`,
+          metadata: {
+            message:
+              'The specified tool was not found. Use search_tools to discover available tools.',
+          },
+        }),
+      },
+    ],
+    isError: true,
+  };
+};
+
+/**
+ * Check if the given group is a smart routing group
+ */
+export const isSmartRoutingGroup = (group: string | undefined): boolean => {
+  return group === '$smart' || (group?.startsWith('$smart/') ?? false);
+};
--- a/src/types/index.ts
+++ b/src/types/index.ts
@@ -270,6 +270,17 @@ export interface McpSettings {
  bearerKeys?: BearerKey[]; // Bearer authentication keys (multi-key configuration)
 }

+// Proxychains4 configuration for STDIO servers (Linux/macOS only)
+export interface ProxychainsConfig {
+  enabled?: boolean; // Enable/disable proxychains4 proxy routing
+  type?: 'socks4' | 'socks5' | 'http'; // Proxy protocol type
+  host?: string; // Proxy server hostname or IP address
+  port?: number; // Proxy server port
+  username?: string; // Proxy authentication username (optional)
+  password?: string; // Proxy authentication password (optional)
+  configPath?: string; // Path to custom proxychains4 configuration file (optional, overrides above settings)
+}
+
 // Configuration details for an individual server
 export interface ServerConfig {
  type?: 'stdio' | 'sse' | 'streamable-http' | 'openapi'; // Type of server
@@ -285,6 +296,8 @@ export interface ServerConfig {
  tools?: Record<string, { enabled: boolean; description?: string }>; // Tool-specific configurations with enable/disable state and custom descriptions
  prompts?: Record<string, { enabled: boolean; description?: string }>; // Prompt-specific configurations with enable/disable state and custom descriptions
  options?: Partial<Pick<RequestOptions, 'timeout' | 'resetTimeoutOnProgress' | 'maxTotalTimeout'>>; // MCP request options configuration
+  // Proxychains4 proxy configuration for STDIO servers (Linux/macOS only, Windows not supported)
+  proxy?: ProxychainsConfig;
  // OAuth authentication for upstream MCP servers
  oauth?: {
    // Static client configuration (traditional OAuth flow)
--- a/src/utils/smartRouting.ts
+++ b/src/utils/smartRouting.ts
@@ -10,6 +10,13 @@ export interface SmartRoutingConfig {
  openaiApiBaseUrl: string;
  openaiApiKey: string;
  openaiApiEmbeddingModel: string;
+  /**
+   * When enabled, search_tools returns only tool name and description (without full inputSchema).
+   * A new describe_tool endpoint is provided to get the full tool schema on demand.
+   * This reduces token usage for AI clients that don't need all tool parameters upfront.
+   * Default: false (returns full tool schemas in search_tools for backward compatibility)
+   */
+  progressiveDisclosure?: boolean;
 }

 /**
@@ -17,7 +24,7 @@ export interface SmartRoutingConfig {
 *
 * Priority order for each setting:
 * 1. Specific environment variables (ENABLE_SMART_ROUTING, SMART_ROUTING_ENABLED, etc.)
- * 2. Generic environment variables (OPENAI_API_KEY, DATABASE_URL, etc.)
+ * 2. Generic environment variables (OPENAI_API_KEY, DB_URL, etc.)
 * 3. Settings configuration (systemConfig.smartRouting)
 * 4. Default values
 *
@@ -62,6 +69,15 @@ export async function getSmartRoutingConfig(): Promise<SmartRoutingConfig> {
      'text-embedding-3-small',
      expandEnvVars,
    ),
+
+    // Progressive disclosure - when enabled, search_tools returns minimal info
+    // and describe_tool is used to get full schema
+    progressiveDisclosure: getConfigValue(
+      [process.env.SMART_ROUTING_PROGRESSIVE_DISCLOSURE],
+      smartRoutingSettings.progressiveDisclosure,
+      false,
+      parseBooleanEnvVar,
+    ),
  };
 }

--- a/tests/services/mcpService-smart-routing-group.test.ts
+++ b/tests/services/mcpService-smart-routing-group.test.ts
@@ -48,6 +48,44 @@ jest.mock('../../src/services/services.js', () => ({
  })),
 }));

+// Mock smartRoutingService to initialize with test functions
+const mockHandleSearchToolsRequest = jest.fn();
+jest.mock('../../src/services/smartRoutingService.js', () => ({
+  initSmartRoutingService: jest.fn(),
+  handleSearchToolsRequest: mockHandleSearchToolsRequest,
+  handleDescribeToolRequest: jest.fn(),
+  isSmartRoutingGroup: jest.fn((group: string) => group?.startsWith('$smart')),
+  getSmartRoutingTools: jest.fn(async (group: string) => {
+    const targetGroup = group?.startsWith('$smart/') ? group.substring(7) : undefined;
+    const scopeDescription = targetGroup
+      ? `servers in the "${targetGroup}" group`
+      : 'all available servers';
+
+    return {
+      tools: [
+        {
+          name: 'search_tools',
+          description: `Search for relevant tools across ${scopeDescription}.`,
+          inputSchema: {
+            type: 'object',
+            properties: { query: { type: 'string' }, limit: { type: 'integer' } },
+            required: ['query'],
+          },
+        },
+        {
+          name: 'call_tool',
+          description: 'Execute a tool by name',
+          inputSchema: {
+            type: 'object',
+            properties: { toolName: { type: 'string' } },
+            required: ['toolName'],
+          },
+        },
+      ],
+    };
+  }),
+}));
+
 jest.mock('../../src/services/vectorSearchService.js', () => ({
  searchToolsByVector: jest.fn(),
  saveToolsAsVectorEmbeddings: jest.fn(),
@@ -66,13 +104,21 @@ jest.mock('../../src/config/index.js', () => ({

 // Import after mocks are set up
 import { handleListToolsRequest, handleCallToolRequest } from '../../src/services/mcpService.js';
-import { getServersInGroup } from '../../src/services/groupService.js';
 import { getGroup } from '../../src/services/sseService.js';
-import { searchToolsByVector } from '../../src/services/vectorSearchService.js';
+import { handleSearchToolsRequest } from '../../src/services/smartRoutingService.js';

 describe('MCP Service - Smart Routing with Group Support', () => {
  beforeEach(() => {
    jest.clearAllMocks();
+    // Setup mock return for handleSearchToolsRequest
+    mockHandleSearchToolsRequest.mockResolvedValue({
+      content: [
+        {
+          type: 'text',
+          text: JSON.stringify({ tools: [], guideline: 'test', nextSteps: 'test' }),
+        },
+      ],
+    });
  });

  describe('handleListToolsRequest', () => {
@@ -89,7 +135,7 @@ describe('MCP Service - Smart Routing with Group Support', () => {
      const result = await handleListToolsRequest({}, { sessionId: 'session-smart-group' });

      expect(getGroup).toHaveBeenCalledWith('session-smart-group');
-      expect(getServersInGroup).toHaveBeenCalledWith('test-group');
+      // Note: getServersInGroup is now called inside the mocked getSmartRoutingTools

      expect(result.tools).toHaveLength(2);
      expect(result.tools[0].name).toBe('search_tools');
@@ -101,7 +147,7 @@ describe('MCP Service - Smart Routing with Group Support', () => {
      const result = await handleListToolsRequest({}, { sessionId: 'session-smart-empty' });

      expect(getGroup).toHaveBeenCalledWith('session-smart-empty');
-      expect(getServersInGroup).toHaveBeenCalledWith('empty-group');
+      // Note: getServersInGroup is now called inside the mocked getSmartRoutingTools

      expect(result.tools).toHaveLength(2);
      expect(result.tools[0].name).toBe('search_tools');
@@ -113,16 +159,6 @@ describe('MCP Service - Smart Routing with Group Support', () => {

  describe('handleCallToolRequest - search_tools', () => {
    it('should search across all servers when using $smart', async () => {
-      const mockSearchResults = [
-        {
-          serverName: 'server1',
-          toolName: 'server1::tool1',
-          description: 'Test tool 1',
-          inputSchema: {},
-        },
-      ];
-      (searchToolsByVector as jest.Mock).mockResolvedValue(mockSearchResults);
-
      const request = {
        params: {
          name: 'search_tools',
@@ -135,25 +171,11 @@ describe('MCP Service - Smart Routing with Group Support', () => {

      await handleCallToolRequest(request, { sessionId: 'session-smart' });

-      expect(searchToolsByVector).toHaveBeenCalledWith(
-        'test query',
-        10,
-        expect.any(Number),
-        undefined, // No server filtering
-      );
+      // handleSearchToolsRequest should be called with the query, limit, and sessionId
+      expect(handleSearchToolsRequest).toHaveBeenCalledWith('test query', 10, 'session-smart');
    });

    it('should filter servers when using $smart/{group}', async () => {
-      const mockSearchResults = [
-        {
-          serverName: 'server1',
-          toolName: 'server1::tool1',
-          description: 'Test tool 1',
-          inputSchema: {},
-        },
-      ];
-      (searchToolsByVector as jest.Mock).mockResolvedValue(mockSearchResults);
-
      const request = {
        params: {
          name: 'search_tools',
@@ -166,20 +188,16 @@ describe('MCP Service - Smart Routing with Group Support', () => {

      await handleCallToolRequest(request, { sessionId: 'session-smart-group' });

-      expect(getGroup).toHaveBeenCalledWith('session-smart-group');
-      expect(getServersInGroup).toHaveBeenCalledWith('test-group');
-      expect(searchToolsByVector).toHaveBeenCalledWith(
+      // handleSearchToolsRequest should be called with the sessionId that contains group info
+      // The group filtering happens inside handleSearchToolsRequest, not in handleCallToolRequest
+      expect(handleSearchToolsRequest).toHaveBeenCalledWith(
        'test query',
        10,
-        expect.any(Number),
-        ['server1', 'server2'], // Filtered to group servers
+        'session-smart-group',
      );
    });

    it('should handle empty group in $smart/{group}', async () => {
-      const mockSearchResults: any[] = [];
-      (searchToolsByVector as jest.Mock).mockResolvedValue(mockSearchResults);
-
      const request = {
        params: {
          name: 'search_tools',
@@ -192,18 +210,19 @@ describe('MCP Service - Smart Routing with Group Support', () => {

      await handleCallToolRequest(request, { sessionId: 'session-smart-empty' });

-      expect(getGroup).toHaveBeenCalledWith('session-smart-empty');
-      expect(getServersInGroup).toHaveBeenCalledWith('empty-group');
-      // Empty group returns empty array, which should still be passed to search
-      expect(searchToolsByVector).toHaveBeenCalledWith(
+      expect(handleSearchToolsRequest).toHaveBeenCalledWith(
        'test query',
        10,
-        expect.any(Number),
-        [], // Empty group
+        'session-smart-empty',
      );
    });

    it('should validate query parameter', async () => {
+      // Mock handleSearchToolsRequest to return an error result when query is missing
+      mockHandleSearchToolsRequest.mockImplementationOnce(() => {
+        return Promise.reject(new Error('Query parameter is required and must be a string'));
+      });
+
      const request = {
        params: {
          name: 'search_tools',
--- a/tests/setup.ts
+++ b/tests/setup.ts
@@ -5,7 +5,7 @@ import 'reflect-metadata';
 Object.assign(process.env, {
  NODE_ENV: 'test',
  JWT_SECRET: 'test-jwt-secret-key',
-  DATABASE_URL: 'sqlite::memory:',
+  DB_URL: 'sqlite::memory:',
 });

 // Mock moduleDir to avoid import.meta parsing issues in Jest
@@ -40,7 +40,7 @@ expect.extend({
      };
    }
  },
-  
+
  toBeValidUUID(received: any) {
    const uuidRegex = /^[0-9a-f]{8}-[0-9a-f]{4}-[1-5][0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}$/i;
    const pass = typeof received === 'string' && uuidRegex.test(received);
Author	SHA1	Message	Date
samanhappy	687fd6448b	fix: update getServerInfos handling to avoid stale references in smart routing service	2026-01-05 13:25:27 +08:00
samanhappy	8008b834ef	feat: add progressive disclosure feature for smart routing tools to reduce token usage	2026-01-05 11:15:38 +08:00
samanhappy	3de56b30bd	fix: update readonlyAllowPaths for correct authorization handling (#550 )	2026-01-04 13:06:27 +08:00
Bryan Thompson	6a08f4bc5a	feat: Add tool annotations for improved LLM tool understanding (#549 ) Co-authored-by: triepod-ai <199543909+triepod-ai@users.noreply.github.com>	2026-01-03 10:20:17 +08:00
samanhappy	ef1bc0d305	fix: add localnet configuration for Proxychains4 (#547 )	2026-01-02 13:42:00 +03:00
samanhappy	b279a1a62c	chore: update mcp sdk dependencies to latest versions (#546 )	2026-01-01 22:41:46 +08:00
dependabot[bot]	760cc462b9	chore(deps-dev): bump tsx from 4.20.5 to 4.21.0 (#541 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2026-01-01 22:01:13 +08:00
dependabot[bot]	431bc8f6f8	chore(deps-dev): bump next from 15.5.9 to 16.1.1 (#543 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2026-01-01 22:00:51 +08:00
dependabot[bot]	fb6af75f5b	chore(deps-dev): bump ts-jest from 29.4.1 to 29.4.6 (#540 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2026-01-01 22:00:17 +08:00
dependabot[bot]	33b440973f	chore(deps): bump axios from 1.13.1 to 1.13.2 (#544 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2026-01-01 21:59:57 +08:00
dependabot[bot]	2d248e953e	chore(deps): bump dotenv from 16.6.1 to 17.2.3 (#542 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2026-01-01 21:59:38 +08:00
samanhappy	d36c6ac5ad	fix: rename DATABASE_URL to DB_URL for consistency across configurations (#545 )	2026-01-01 21:58:11 +08:00
Zhyim	0be6c36e12	feat: implement pagination for server list with customizable items pe… (#534 )	2026-01-01 13:36:09 +08:00
samanhappy	7f2fca9636	feat: add proxy configuration support for STDIO servers on Linux and macOS (#537 )	2026-01-01 12:45:50 +08:00