YingDao RPA MCP Server

https://github.com/ying-dao/yingdao_mcp_server
Admin
An MCP server for YingDao RPA that integrates RPA automation capabilities with AI tools like Claude...
Tokens:9,774
Snippets:50
Trust Score:4.9
License:MIT
Update:6 months ago
Show doc for...
Tokens:
Raw
### Build and Start SSE Server

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Commands to clone, build, and start the SSE server.

```bash
git clone https://github.com/ying-dao/yingdao_mcp_server.git
cd yingdao_mcp_server
npm install
npm run build
```

```bash
npm run start:server
```

--------------------------------

### Server Setup and Configuration

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Instructions for setting up and configuring the 影刀 RPA MCP Server in different modes.

```APIDOC
## Server Setup and Configuration

### Stdio Mode Setup

Stdio mode is suitable for local integration with clients like Claude Desktop and Cursor. It can be run directly using npx without manual server deployment.

**MCP Client Configuration (e.g., `claude_desktop_config.json`)**
```json
{
  "mcpServers": {
    "YingDao RPA MCP Server": {
      "command": "npx",
      "args": ["-y", "yingdao-mcp-server"],
      "env": {
        "RPA_MODEL": "openApi",
        "ACCESS_KEY_ID": "your_access_key_id",
        "ACCESS_KEY_SECRET": "your_access_key_secret"
      }
    }
  }
}
```

### SSE Server Mode Setup

SSE mode provides an HTTP server for remote client connections, ideal for server deployments accessible by multiple clients.

**Project Setup**
```bash
git clone https://github.com/ying-dao/yingdao_mcp_server.git
cd yingdao_mcp_server
npm install
npm run build
```

**Environment Configuration (.env file)**
```
RPA_MODEL=openApi
ACCESS_KEY_ID=your_access_key_id
ACCESS_KEY_SECRET=your_access_key_secret
SERVER_PORT=3000
```

**Start SSE Server**
```bash
npm run start:server
# Output: RPA HTTP Server listening on port 3000
# SSE endpoint: http://localhost:3000/sse
# Message endpoint: http://localhost:3000/messages
```

**AI Power Client SSE Configuration**
```json
{
  "mcpServers": {
    "YingDao RPA MCP Server": {
      "url": "http://localhost:3000/sse",
      "description": "影刀 MCP Server"
    }
  }
}
```

### Local Mode Configuration

Local mode directly utilizes the locally installed 影刀 RPA client, without requiring cloud API keys.

**Local Mode Environment Configuration (.env file)**
```
RPA_MODEL=local
LANGUAGE=zh
```

**Windows Path Configuration (use double backslashes)**
```
SHADOWBOT_PATH=D:\\Program Files\\YingDao\\ShadowBot.exe
USER_FOLDER=C:\\Users\\YourName\\YingDaoRPA
```

**Mac Path Configuration**
```
SHADOWBOT_PATH=/Applications/影刀.app
USER_FOLDER=/Users/YourName/YingDaoRPA
```
```

--------------------------------

### queryApplist - 查询本地 RPA 应用列表 (本地模式)

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this API scans the user's folder for installed RPA applications and returns basic application information. It only returns 'My Apps' that have been executed at least once.

```APIDOC
## POST /api/queryApplist (Local Mode)

### Description
In local mode, this API scans the user's folder for installed RPA applications and returns basic application information. It only returns 'My Apps' that have been executed at least once.

### Method
POST

### Endpoint
/api/queryApplist

### Parameters
#### Request Body
(No parameters required for this endpoint)

### Request Example
```json
{
}
```

### Response
#### Success Response (200)
- **uuid** (string) - The UUID of the RPA application.
- **name** (string) - The name of the RPA application.
- **description** (string) - A brief description of the application.

#### Response Example
```json
[
  {
    "uuid": "local-app-uuid-001",
    "name": "日报自动生成",
    "description": "每日自动汇总数据并生成日报"
  },
  {
    "uuid": "local-app-uuid-002",
    "name": "邮件自动分类",
    "description": "自动分类收件箱邮件到对应文件夹"
  }
]
```
```

--------------------------------

### Build and Run SSE Server Mode

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Steps to clone, build, configure environment variables, and start the YingDao RPA MCP Server in SSE mode. This mode provides an HTTP server for remote client access.

```bash
# 克隆并构建项目
git clone https://github.com/ying-dao/yingdao_mcp_server.git
cd yingdao_mcp_server
npm install
npm run build

# 配置环境变量 (.env 文件)
RPA_MODEL=openApi
ACCESS_KEY_ID=your_access_key_id
ACCESS_KEY_SECRET=your_access_key_secret
SERVER_PORT=3000

# 启动 SSE 服务器
npm run start:server
# 输出: RPA HTTP Server listening on port 3000
# SSE endpoint: http://localhost:3000/sse
# Message endpoint: http://localhost:3000/messages
```

--------------------------------

### startJob - 启动 RPA 作业

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Starts an RPA application job. You can specify a robot or a robot group for execution and pass runtime parameters. Either accountName or robotClientGroupUuid must be provided to specify the executing robot.

```APIDOC
## POST /api/startJob

### Description
Starts an RPA application job. You can specify a robot or a robot group for execution and pass runtime parameters. Either accountName or robotClientGroupUuid must be provided to specify the executing robot.

### Method
POST

### Endpoint
/api/startJob

### Parameters
#### Request Body
- **robotUuid** (string) - Required - RPA application UUID
- **accountName** (string) - Optional - Robot account name (must be in 'idle' state). Mutually exclusive with robotClientGroupUuid.
- **robotClientGroupUuid** (string) - Optional - Robot group UUID. Mutually exclusive with accountName.
- **waitTimeoutSeconds** (integer) - Optional - Waiting timeout in seconds (range: 60-950400).
- **runTimeout** (integer) - Optional - Runtime timeout in seconds (range: 60-950400).
- **params** (object) - Optional - Runtime parameters.
  - **invoicePath** (string) - Example parameter for job.
  - **outputFormat** (string) - Example parameter for job.
  - **fileKey** (string) - Example parameter for job.

### Request Example
```json
{
  "robotUuid": "robot-uuid-abc123",
  "accountName": "robot-client-001",
  "waitTimeoutSeconds": 300,
  "runTimeout": 600,
  "params": {
    "invoicePath": "/data/invoices/batch001",
    "outputFormat": "json",
    "fileKey": "file-key-xyz789"
  }
}
```

### Response
#### Success Response (200)
- **code** (integer) - Response code.
- **success** (boolean) - Indicates if the operation was successful.
- **msg** (string) - Message describing the result.
- **data** (object) - Contains job details.
  - **jobUuid** (string) - The UUID of the created job.
  - **idempotentFlag** (boolean) - Flag indicating idempotency.

#### Response Example
```json
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "jobUuid": "job-uuid-xyz123",
    "idempotentFlag": false
  }
}
```
```

--------------------------------

### queryClientList - 查询机器人列表

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Queries the list of schedulable RPA robot clients. Supports filtering by status and robot group. Used to select appropriate robots before starting a job.

```APIDOC
## POST /api/queryClientList

### Description
Queries the list of schedulable RPA robot clients. Supports filtering by status and robot group. Used to select appropriate robots before starting a job.

### Method
POST

### Endpoint
/api/queryClientList

### Parameters
#### Request Body
- **status** (string) - Optional - Filter by status (e.g., 'idle').
- **key** (string) - Optional - Keyword search.
- **robotClientGroupUuid** (string) - Optional - Filter by robot group UUID.
- **page** (string) - Optional - Page number (default: 1).
- **size** (string) - Optional - Number of items per page (default: 30, max: 100).

### Request Example
```json
{
  "status": "idle",
  "key": "production",
  "robotClientGroupUuid": "group-uuid",
  "page": "1",
  "size": "30"
}
```

### Response
#### Success Response (200)
- **code** (integer) - Response code.
- **success** (boolean) - Indicates if the operation was successful.
- **msg** (string) - Message describing the result.
- **data** (array) - List of robot clients.
  - **robotClientUuid** (string) - UUID of the robot client.
  - **robotClientName** (string) - Name of the robot client.
  - **status** (string) - Current status of the robot client.
  - **description** (string) - Description of the robot client.
  - **windowsUserName** (string) - Windows username associated with the client.
  - **clientIp** (string) - IP address of the robot client.
  - **machineName** (string) - Hostname of the machine.
- **page** (object) - Pagination details.
  - **total** (integer) - Total number of items.
  - **size** (integer) - Items per page.
  - **page** (integer) - Current page number.
  - **pages** (integer) - Total number of pages.

#### Response Example
```json
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": [
    {
      "robotClientUuid": "client-uuid-001",
      "robotClientName": "production-robot-01",
      "status": "idle",
      "description": "生产环境机器人1",
      "windowsUserName": "RobotUser",
      "clientIp": "192.168.1.101",
      "machineName": "PROD-SERVER-01"
    },
    {
      "robotClientUuid": "client-uuid-002",
      "robotClientName": "production-robot-02",
      "status": "idle",
      "description": "生产环境机器人2",
      "windowsUserName": "RobotUser",
      "clientIp": "192.168.1.102",
      "machineName": "PROD-SERVER-02"
    }
  ],
  "page": {
    "total": 5,
    "size": 30,
    "page": 1,
    "pages": 1
  }
}
```
```

--------------------------------

### Configure Local Mode Environment Variables

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Environment variable setup for the YingDao RPA MCP Server when running in local mode. This mode directly uses the local RPA client and requires specifying paths.

```bash
# 本地模式环境变量配置 (.env 文件)
RPA_MODEL=local
LANGUAGE=zh

# Windows 路径配置 (注意使用双斜杠)
SHADOWBOT_PATH=D://Program Files//YingDao//ShadowBot.exe
USER_FOLDER=C://Users//YourName//YingDaoRPA

# Mac 路径配置
SHADOWBOT_PATH=/Applications/影刀.app
USER_FOLDER=/Users/YourName/YingDaoRPA
```

--------------------------------

### Query Local RPA Application List

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this scans the user's folder for installed RPA applications and returns their basic information. Only returns applications that have been executed at least once.

```typescript
const result = await mcpClient.callTool('queryApplist', {});

// 返回结果示例
[
  {
    "uuid": "local-app-uuid-001",
    "name": "日报自动生成",
    "description": "每日自动汇总数据并生成日报"
  },
  {
    "uuid": "local-app-uuid-002",
    "name": "邮件自动分类",
    "description": "自动分类收件箱邮件到对应文件夹"
  }
]
```

--------------------------------

### Start RPA Job

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Initiates an RPA application job. Specify either robot account name or group UUID. Supports passing runtime parameters and setting timeouts.

```typescript
const result = await mcpClient.callTool('startJob', {
  robotUuid: 'robot-uuid-abc123',        // 必填：RPA 应用 UUID
  accountName: 'robot-client-001',       // 可选：机器人账号名称 (需状态为 idle)
  // robotClientGroupUuid: 'group-uuid', // 可选：机器人组 UUID (与 accountName 互斥)
  waitTimeoutSeconds: 300,               // 可选：等待超时时间(秒)，范围 60-950400
  runTimeout: 600,                       // 可选：运行超时时间(秒)，范围 60-950400
  params: {                              // 可选：运行参数
    invoicePath: '/data/invoices/batch001',
    outputFormat: 'json',
    fileKey: 'file-key-xyz789'
  }
});

// 返回结果示例
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "jobUuid": "job-uuid-xyz123",
    "idempotentFlag": false
  }
}
```

--------------------------------

### Query RPA Robot Client List

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Fetches a list of available RPA robot clients. Supports filtering by status, keyword, and robot group. Useful for selecting an execution robot before starting a job.

```typescript
const result = await mcpClient.callTool('queryClientList', {
  status: 'idle',
  key: 'production',
  robotClientGroupUuid: 'group-uuid',
  page: '1',
  size: '30'
});

// 返回结果示例
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": [
    {
      "robotClientUuid": "client-uuid-001",
      "robotClientName": "production-robot-01",
      "status": "idle",
      "description": "生产环境机器人1",
      "windowsUserName": "RobotUser",
      "clientIp": "192.168.1.101",
      "machineName": "PROD-SERVER-01"
    },
    {
      "robotClientUuid": "client-uuid-002",
      "robotClientName": "production-robot-02",
      "status": "idle",
      "description": "生产环境机器人2",
      "windowsUserName": "RobotUser",
      "clientIp": "192.168.1.102",
      "machineName": "PROD-SERVER-02"
    }
  ],
  "page": {
    "total": 5,
    "size": 30,
    "page": 1,
    "pages": 1
  }
}
```

--------------------------------

### Configure Stdio Server in Client

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Add this configuration to your MCP host client to run the server via Stdio.

```json
{
  "mcpServers": {
    "YingDao RPA MCP Server": {
      "command": "npx",
      "args": ["-y", "yingdao-mcp-server"],
      "env":{
        "RPA_MODEL":"openApi",
        "ACCESS_KEY_ID":"{your_access_key_id}",
        "ACCESS_KEY_SECRET":"{your_access_key_secret}"
      }
    }
  }
}
```

--------------------------------

### Configure SSE Server in Client

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Add this configuration to your AI Power client to connect to the SSE server.

```json
{
  "mcpServers": {
    "YingDao RPA MCP Server": {
      "url": "http://localhost:3000/sse",
      "description": "影刀 MCP Server"
    }
  }
}
```

--------------------------------

### runApp - 运行本地 RPA 应用 (本地模式)

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this API directly launches the local Yingdao RPA client to execute a specified application. It uses the 'shadowbot:' protocol to call the local client.

```APIDOC
## POST /api/runApp (Local Mode)

### Description
In local mode, this API directly launches the local Yingdao RPA client to execute a specified application. It uses the 'shadowbot:' protocol to call the local client.

### Method
POST

### Endpoint
/api/runApp

### Parameters
#### Request Body
- **appUuid** (string) - Required - The UUID of the application.
- **appParams** (object) - Optional - Parameters for the application.
  - **reportDate** (string) - Example parameter for the application.
  - **outputPath** (string) - Example parameter for the application.

### Request Example
```json
{
  "appUuid": "local-app-uuid-001",
  "appParams": {
    "reportDate": "2024-01-25",
    "outputPath": "/Users/name/Reports"
  }
}
```

### Response
#### Success Response (200)
- **result** (string) - Indicates success, typically returns "success".

#### Response Example
```json
"success"
```

### Notes
Internal execution command examples:
- Windows: `spawn("D://Program Files//YingDao//ShadowBot.exe", ["shadowbot:Run?robot-uuid=local-app-uuid-001"])
- Mac: `spawn("/Applications/影刀.app/Contents/MacOS/影刀", ["shadowbot:Run?robot-uuid=local-app-uuid-001"])
```
```

--------------------------------

### Configure Open API Mode Environment Variables

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Set these variables for enterprise users to authenticate via Open API.

```bash
RPA_MODEL=openApi
ACCESS_KEY_ID={your_access_key_id}
ACCESS_KEY_SECRET={your_access_key_secret}
```

--------------------------------

### Run Local RPA Application

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this directly launches the local Yingdao RPA client to execute a specified application using the shadowbot: protocol.

```typescript
const result = await mcpClient.callTool('runApp', {
  appUuid: 'local-app-uuid-001',
  appParams: {
    reportDate: '2024-01-25',
    outputPath: '/Users/name/Reports'
  }
});

// 返回结果
"success"
```

```shell
// 内部执行命令 (Windows)
// spawn("D://Program Files//YingDao//ShadowBot.exe", ["shadowbot:Run?robot-uuid=local-app-uuid-001"])

// 内部执行命令 (Mac)
// spawn("/Applications/影刀.app/Contents/MacOS/影刀", ["shadowbot:Run?robot-uuid=local-app-uuid-001"])
```

--------------------------------

### Configure Local Mode Environment Variables

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Set these environment variables to enable local RPA execution. Ensure the application has been executed at least once.

```bash
RPA_MODEL=local
SHADOWBOT_PATH={your_shadowbot_path} //影刀rpa的exe路径
USER_FOLDER={your_user_folder}       //影刀rpa的用户文件夹路径
```

--------------------------------

### Query RPA Application List

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Use the 'queryApplist' tool to retrieve a paginated list of RPA applications. Supports filtering by name, ID, and owner. Ensure 'mcpClient' is initialized.

```typescript
// MCP Tool 调用示例
const result = await mcpClient.callTool('queryApplist', {
  appName: '发票处理',      // 可选：模糊匹配应用名称
  appId: 'app-uuid-123',   // 可选：精确匹配应用 ID
  ownerUserSearchKey: 'admin@company.com', // 可选：所有者账号精确匹配
  page: '1',               // 页码，默认 1
  size: '30'               // 每页数量，默认 30，最大 100
});

// 返回结果示例
{
  "total": 15,
  "list": [
    {
      "id": "robot-uuid-abc123",
      "name": "发票识别自动化",
      "ownerUser": "admin@company.com",
      "createTime": "2024-01-15 10:30:00"
    },
    {
      "id": "robot-uuid-def456",
      "name": "发票归档处理",
      "ownerUser": "admin@company.com",
      "createTime": "2024-01-20 14:15:00"
    }
  ]
}
```

--------------------------------

### Configure MCP Client for Stdio Mode

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Configuration for MCP clients like Claude Desktop to use the YingDao RPA MCP Server in Stdio mode via npx. Ensure 'RPA_MODEL' is set to 'openApi' and provide access keys.

```json
{
  "mcpServers": {
    "YingDao RPA MCP Server": {
      "command": "npx",
      "args": ["-y", "yingdao-mcp-server"],
      "env": {
        "RPA_MODEL": "openApi",
        "ACCESS_KEY_ID": "your_access_key_id",
        "ACCESS_KEY_SECRET": "your_access_key_secret"
      }
    }
  }
}
```

--------------------------------

### Define ShadowBot Executable Path

Source: https://github.com/ying-dao/yingdao_mcp_server/blob/main/README.md

Specify the path to the ShadowBot executable for Windows or Mac environments.

```bash
D://Program Files//{安装目录}//ShadowBot.exe
```

```bash
/Applications/影刀.app
```

--------------------------------

### Query Local Application Parameters

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this reads the application's flow configuration file to obtain parameter definitions.

```typescript
const result = await mcpClient.callTool('queryRobotParam', {
  robotUuid: 'local-app-uuid-001'
});

// 返回结果示例
[
  {
    "name": "reportDate",
    "type": "string",
    "description": "报表日期"
  },
  {
    "name": "outputPath",
    "type": "string",
    "description": "输出目录路径"
  }
]
```

--------------------------------

### Query RPA Application Parameters

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Use the 'queryRobotParam' tool to fetch input and output parameter definitions for an RPA application. Can query by UUID or exact name. Ensure 'mcpClient' is initialized.

```typescript
// MCP Tool 调用示例
const result = await mcpClient.callTool('queryRobotParam', {
  robotUuid: 'robot-uuid-abc123',     // 可选：RPA 应用 UUID
  accurateRobotName: '发票识别自动化'   // 可选：精确匹配的应用名称
});

// 返回结果示例
{
  "inputParams": [
    {
      "name": "invoicePath",
      "direction": "input",
      "type": "string",
      "value": "",
      "description": "发票文件路径",
      "kind": "variable"
    },
    {
      "name": "outputFormat",
      "direction": "input",
      "type": "string",
      "value": "json",
      "description": "输出格式类型",
      "kind": "variable"
    }
  ],
  "outputParams": [
    {
      "name": "invoiceData",
      "direction": "output",
      "type": "string",
      "value": "",
      "description": "识别后的发票数据",
      "kind": "variable"
    }
  ]
}
```

--------------------------------

### Upload File to RPA Platform

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Use the 'uploadFile' tool to upload files (txt, csv, xlsx) to the RPA platform. Returns a file key for subsequent use. File names are limited to 100 characters and size to 10MB. Ensure 'mcpClient' is initialized.

```typescript
// MCP Tool 调用示例
const fileContent = Buffer.from('姓名,部门,工号\n张三,技术部,001\n李四,财务部,002');
const result = await mcpClient.callTool('uploadFile', {
  file: fileContent,          // 文件内容 (Buffer 或字符串)
  fileName: 'employees.csv'   // 文件名，支持 txt/csv/xlsx
});

// 返回结果示例
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "fileKey": "file-key-xyz789"
  }
}

// 错误处理示例
try {
  await mcpClient.callTool('uploadFile', {
    file: largeContent,
    fileName: 'very_long_filename_that_exceeds_one_hundred_characters_limit_should_fail.xlsx'
  });
} catch (error) {
  // 错误: 文件名长度不能超过100
}
```

--------------------------------

### queryRobotParam - 查询本地应用参数 (本地模式)

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

In local mode, this API reads the application's flow configuration file to obtain parameter definitions.

```APIDOC
## POST /api/queryRobotParam (Local Mode)

### Description
In local mode, this API reads the application's flow configuration file to obtain parameter definitions.

### Method
POST

### Endpoint
/api/queryRobotParam

### Parameters
#### Request Body
- **robotUuid** (string) - Optional - The UUID of the application.

### Request Example
```json
{
  "robotUuid": "local-app-uuid-001"
}
```

### Response
#### Success Response (200)
- **name** (string) - The name of the parameter.
- **type** (string) - The data type of the parameter.
- **description** (string) - A description of the parameter.

#### Response Example
```json
[
  {
    "name": "reportDate",
    "type": "string",
    "description": "报表日期"
  },
  {
    "name": "outputPath",
    "type": "string",
    "description": "输出目录路径"
  }
]
```
```

--------------------------------

### 自动化发票处理工作流

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

展示从查询应用、上传文件到启动 RPA 作业并轮询结果的完整流程。

```typescript
// 完整的 AI 自动化工作流示例
async function processInvoices(mcpClient, invoiceFile) {
  // 1. 查询可用的 RPA 应用
  const apps = await mcpClient.callTool('queryApplist', {
    appName: '发票'
  });
  console.log('找到应用:', apps.list.length);

  const invoiceApp = apps.list.find(app => app.name.includes('发票识别'));
  if (!invoiceApp) throw new Error('未找到发票识别应用');

  // 2. 查询应用参数
  const params = await mcpClient.callTool('queryRobotParam', {
    robotUuid: invoiceApp.id
  });
  console.log('应用参数:', params.inputParams);

  // 3. 上传待处理文件
  const uploadResult = await mcpClient.callTool('uploadFile', {
    file: invoiceFile,
    fileName: 'invoices.xlsx'
  });
  const fileKey = uploadResult.data.fileKey;

  // 4. 查询空闲机器人
  const clients = await mcpClient.callTool('queryClientList', {
    status: 'idle',
    size: '10'
  });

  if (clients.data.length === 0) {
    throw new Error('没有可用的空闲机器人');
  }
  const selectedClient = clients.data[0];

  // 5. 启动 RPA 作业
  const job = await mcpClient.callTool('startJob', {
    robotUuid: invoiceApp.id,
    accountName: selectedClient.robotClientName,
    waitTimeoutSeconds: 300,
    runTimeout: 1800,
    params: {
      inputFileKey: fileKey,
      outputFormat: 'json'
    }
  });
  console.log('作业已启动:', job.data.jobUuid);

  // 6. 轮询等待作业完成
  let jobResult;
  for (let i = 0; i < 120; i++) {
    jobResult = await mcpClient.callTool('queryJob', {
      jobUuid: job.data.jobUuid
    });

    if (jobResult.data.status === 'success') {
      console.log('作业执行成功!');
      console.log('输出结果:', jobResult.data.robotParams?.outputs);
      return jobResult.data;
    } else if (jobResult.data.status === 'failed') {
      throw new Error(`作业执行失败: ${jobResult.data.remark}`);
    }

    await new Promise(r => setTimeout(r, 5000));
  }

  throw new Error('作业执行超时');
}
```

--------------------------------

### MCP Tools - Open API Mode

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

APIs available in the Open API mode for interacting with RPA applications.

```APIDOC
## MCP Tools - Open API Mode

### queryApplist - Query RPA Application List

This interface is used to paginate and retrieve a list of RPA applications, supporting filtering by application name, application ID, and owner. The results include basic application information such as name, ID, and creation time.

**Method**
`callTool('queryApplist', { ... })`

**Parameters**
- **appName** (string, optional) - Fuzzy match application name.
- **appId** (string, optional) - Exact match application ID.
- **ownerUserSearchKey** (string, optional) - Exact match for the owner's account.
- **page** (string, optional) - Page number, defaults to '1'.
- **size** (string, optional) - Number of items per page, defaults to '30', maximum '100'.

**Request Example**
```typescript
const result = await mcpClient.callTool('queryApplist', {
  appName: '发票处理',
  appId: 'app-uuid-123',
  ownerUserSearchKey: 'admin@company.com',
  page: '1',
  size: '30'
});
```

**Response Example (Success)**
```json
{
  "total": 15,
  "list": [
    {
      "id": "robot-uuid-abc123",
      "name": "发票识别自动化",
      "ownerUser": "admin@company.com",
      "createTime": "2024-01-15 10:30:00"
    },
    {
      "id": "robot-uuid-def456",
      "name": "发票归档处理",
      "ownerUser": "admin@company.com",
      "createTime": "2024-01-20 14:15:00"
    }
  ]
}
```

### queryRobotParam - Query RPA Application Parameters

This interface queries the input and output parameter definitions for an RPA application, which is useful for obtaining required parameter information before launching an application. It supports querying by UUID or exact name.

**Method**
`callTool('queryRobotParam', { ... })`

**Parameters**
- **robotUuid** (string, optional) - UUID of the RPA application.
- **accurateRobotName** (string, optional) - Exact match for the application name.

**Request Example**
```typescript
const result = await mcpClient.callTool('queryRobotParam', {
  robotUuid: 'robot-uuid-abc123',
  accurateRobotName: '发票识别自动化'
});
```

**Response Example (Success)**
```json
{
  "inputParams": [
    {
      "name": "invoicePath",
      "direction": "input",
      "type": "string",
      "value": "",
      "description": "发票文件路径",
      "kind": "variable"
    },
    {
      "name": "outputFormat",
      "direction": "input",
      "type": "string",
      "value": "json",
      "description": "输出格式类型",
      "kind": "variable"
    }
  ],
  "outputParams": [
    {
      "name": "invoiceData",
      "direction": "output",
      "type": "string",
      "value": "",
      "description": "识别后的发票数据",
      "kind": "variable"
    }
  ]
}
```

### uploadFile - Upload File to RPA Platform

This interface uploads a file to the RPA platform and returns a file key for subsequent use by RPA applications. It supports txt, csv, and xlsx formats. Filenames should not exceed 100 characters, and file size is limited to 10MB.

**Method**
`callTool('uploadFile', { ... })`

**Parameters**
- **file** (Buffer or string) - File content.
- **fileName** (string) - Filename, supports txt/csv/xlsx.

**Request Example**
```typescript
const fileContent = Buffer.from('姓名,部门,工号\n张三,技术部,001\n李四,财务部,002');
const result = await mcpClient.callTool('uploadFile', {
  file: fileContent,
  fileName: 'employees.csv'
});
```

**Response Example (Success)**
```json
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "fileKey": "file-key-xyz789"
  }
}
```

**Error Handling Example**
```typescript
try {
  await mcpClient.callTool('uploadFile', {
    file: largeContent,
    fileName: 'very_long_filename_that_exceeds_one_hundred_characters_limit_should_fail.xlsx'
  });
} catch (error) {
  // Error: Filename length cannot exceed 100 characters
}
```
```

--------------------------------

### TokenManager 自动令牌管理

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

实现访问令牌的自动获取、缓存与过期前刷新逻辑。

```typescript
// TokenManager 内部实现逻辑
class TokenManager {
  private token: string | null = null;
  private tokenExpireTime: number = 0;

  constructor(accessKeyId: string, accessKeySecret: string) {
    // 初始化认证信息
  }

  async getToken(): Promise<string> {
    // 检查缓存的 token 是否有效
    if (this.token && !this.isTokenExpired()) {
      return this.token;
    }

    // 获取新 token
    const response = await this.client.get('/oapi/token/v2/token/create', {
      params: {
        accessKeyId: this.accessKeyId,
        accessKeySecret: this.accessKeySecret,
      },
    });

    this.token = response.data.data.accessToken;
    // 提前 5 分钟刷新
    this.tokenExpireTime = Date.now() + (response.data.data.expiresIn * 1000) - 300000;

    return this.token;
  }
}
```

```json
{
  "code": 200,
  "success": true,
  "requestId": "req-123456",
  "data": {
    "accessToken": "eyJhbGciOiJIUzI1NiIs...",
    "expiresIn": 7200  // 有效期(秒)
  }
}
```

--------------------------------

### queryJob - 查询作业状态

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Queries the execution status and results of an RPA job, including runtime status, execution time, and output parameters. Useful for polling job completion status.

```APIDOC
## POST /api/queryJob

### Description
Queries the execution status and results of an RPA job, including runtime status, execution time, and output parameters. Useful for polling job completion status.

### Method
POST

### Endpoint
/api/queryJob

### Parameters
#### Request Body
- **jobUuid** (string) - Required - The UUID of the job.

### Request Example
```json
{
  "jobUuid": "job-uuid-xyz123"
}
```

### Response
#### Success Response (200)
- **code** (integer) - Response code.
- **success** (boolean) - Indicates if the operation was successful.
- **msg** (string) - Message describing the result.
- **data** (object) - Contains job details.
  - **jobUuid** (string) - The UUID of the job.
  - **status** (string) - The current status of the job (e.g., 'success', 'failed', 'running').
  - **statusName** (string) - A human-readable name for the job status.
  - **robotUuid** (string) - The UUID of the robot that executed the job.
  - **robotName** (string) - The name of the robot.
  - **robotClientUuid** (string) - The UUID of the robot client.
  - **robotClientName** (string) - The name of the robot client.
  - **startTime** (string) - The start time of the job execution.
  - **endTime** (string) - The end time of the job execution.
  - **robotParams** (object) - Parameters used during robot execution.
    - **inputs** (array) - Input parameters.
      - **name** (string) - Parameter name.
      - **value** (string) - Parameter value.
      - **type** (string) - Parameter type.
    - **outputs** (array) - Output parameters.
      - **name** (string) - Parameter name.
      - **value** (string) - Parameter value.
      - **type** (string) - Parameter type.

#### Response Example
```json
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "jobUuid": "job-uuid-xyz123",
    "status": "success",
    "statusName": "执行成功",
    "robotUuid": "robot-uuid-abc123",
    "robotName": "发票识别自动化",
    "robotClientUuid": "client-uuid-001",
    "robotClientName": "robot-client-001",
    "startTime": "2024-01-25 10:30:00",
    "endTime": "2024-01-25 10:32:15",
    "robotParams": {
      "inputs": [
        { "name": "invoicePath", "value": "/data/invoices/batch001", "type": "string" }
      ],
      "outputs": [
        { "name": "invoiceData", "value": "{\"total\": 25, \"processed\": 25}", "type": "string" }
      ]
    }
  }
}
```
```

--------------------------------

### Query RPA Job Status

Source: https://context7.com/ying-dao/yingdao_mcp_server/llms.txt

Retrieves the execution status and results of an RPA job. Useful for polling job completion. Handles success, failure, and cancellation states.

```typescript
const result = await mcpClient.callTool('queryJob', {
  jobUuid: 'job-uuid-xyz123'
});

// 返回结果示例
{
  "code": 200,
  "success": true,
  "msg": "success",
  "data": {
    "jobUuid": "job-uuid-xyz123",
    "status": "success",
    "statusName": "执行成功",
    "robotUuid": "robot-uuid-abc123",
    "robotName": "发票识别自动化",
    "robotClientUuid": "client-uuid-001",
    "robotClientName": "robot-client-001",
    "startTime": "2024-01-25 10:30:00",
    "endTime": "2024-01-25 10:32:15",
    "robotParams": {
      "inputs": [
        { "name": "invoicePath", "value": "/data/invoices/batch001", "type": "string" }
      ],
      "outputs": [
        { "name": "invoiceData", "value": "{\"total\": 25, \"processed\": 25}", "type": "string" }
      ]
    }
  }
}
```

```typescript
async function waitForJobCompletion(mcpClient, jobUuid, maxAttempts = 60) {
  for (let i = 0; i < maxAttempts; i++) {
    const result = await mcpClient.callTool('queryJob', { jobUuid });
    const status = result.data.status;

    if (status === 'success') {
      return { success: true, data: result.data };
    } else if (status === 'failed' || status === 'canceled') {
      return { success: false, data: result.data };
    }

    await new Promise(resolve => setTimeout(resolve, 5000)); // 等待 5 秒
  }
  throw new Error('作业执行超时');
}
```
Tokens:
Raw